This commit is contained in:
commit
fc460a740b
|
|
@ -0,0 +1,734 @@
|
|||
import cv2,os,time,json
|
||||
from models.experimental import attempt_load
|
||||
from segutils.segmodel import SegModel,get_largest_contours
|
||||
from segutils.trtUtils import segtrtEval,yolov5Trtforward,OcrTrtForward
|
||||
from segutils.trafficUtils import tracfficAccidentMixFunction
|
||||
|
||||
|
||||
from utils.torch_utils import select_device
|
||||
from utilsK.queRiver import get_labelnames,get_label_arrays,post_process_,img_pad,draw_painting_joint,detectDraw,getDetections,getDetectionsFromPreds
|
||||
from trackUtils.sort import moving_average_wang
|
||||
|
||||
from utils.datasets import letterbox
|
||||
import numpy as np
|
||||
import torch
|
||||
import math
|
||||
from PIL import Image
|
||||
import torch.nn.functional as F
|
||||
from copy import deepcopy
|
||||
from scipy import interpolate
|
||||
import glob
|
||||
|
||||
from loguru import logger
|
||||
|
||||
def get_images_videos(impth, imageFixs=['.jpg','.JPG','.PNG','.png'],videoFixs=['.MP4','.mp4','.avi']):
|
||||
imgpaths=[];###获取文件里所有的图像
|
||||
videopaths=[]###获取文件里所有的视频
|
||||
if os.path.isdir(impth):
|
||||
for postfix in imageFixs:
|
||||
imgpaths.extend(glob.glob('%s/*%s'%(impth,postfix )) )
|
||||
for postfix in videoFixs:
|
||||
videopaths.extend(glob.glob('%s/*%s'%(impth,postfix )) )
|
||||
else:
|
||||
postfix = os.path.splitext(impth)[-1]
|
||||
if postfix in imageFixs: imgpaths=[ impth ]
|
||||
if postfix in videoFixs: videopaths = [impth ]
|
||||
|
||||
print('%s: test Images:%d , test videos:%d '%(impth, len(imgpaths), len(videopaths)))
|
||||
return imgpaths,videopaths
|
||||
|
||||
|
||||
def xywh2xyxy(box,iW=None,iH=None):
|
||||
xc,yc,w,h = box[0:4]
|
||||
x0 =max(0, xc-w/2.0)
|
||||
x1 =min(1, xc+w/2.0)
|
||||
y0=max(0, yc-h/2.0)
|
||||
y1=min(1,yc+h/2.0)
|
||||
if iW: x0,x1 = x0*iW,x1*iW
|
||||
if iH: y0,y1 = y0*iH,y1*iH
|
||||
return [x0,y0,x1,y1]
|
||||
|
||||
|
||||
def get_ms(t2,t1):
|
||||
return (t2-t1)*1000.0
|
||||
def get_postProcess_para(parfile):
|
||||
with open(parfile) as fp:
|
||||
par = json.load(fp)
|
||||
assert 'post_process' in par.keys(), ' parfile has not key word:post_process'
|
||||
parPost=par['post_process']
|
||||
|
||||
return parPost["conf_thres"],parPost["iou_thres"],parPost["classes"],parPost["rainbows"]
|
||||
def get_postProcess_para_dic(parfile):
|
||||
with open(parfile) as fp:
|
||||
par = json.load(fp)
|
||||
parPost=par['post_process']
|
||||
return parPost
|
||||
def score_filter_byClass(pdetections,score_para_2nd):
|
||||
ret=[]
|
||||
for det in pdetections:
|
||||
score,cls = det[4],det[5]
|
||||
if int(cls) in score_para_2nd.keys():
|
||||
score_th = score_para_2nd[int(cls)]
|
||||
elif str(int(cls)) in score_para_2nd.keys():
|
||||
score_th = score_para_2nd[str(int(cls))]
|
||||
else:
|
||||
score_th = 0.7
|
||||
if score > score_th:
|
||||
ret.append(det)
|
||||
return ret
|
||||
|
||||
def AI_process(im0s,model,segmodel,names,label_arraylist,rainbows,objectPar={ 'half':True,'device':'cuda:0' ,'conf_thres':0.25,'iou_thres':0.45,'allowedList':[0,1,2,3],'segRegionCnt':1, 'trtFlag_det':False,'trtFlag_seg':False,'score_byClass':{x:0.1 for x in range(30)} }, font={ 'line_thickness':None, 'fontSize':None,'boxLine_thickness':None,'waterLineColor':(0,255,255),'waterLineWidth':3} ,segPar={'modelSize':(640,360),'mean':(0.485, 0.456, 0.406),'std' :(0.229, 0.224, 0.225),'numpy':False, 'RGB_convert_first':True},mode='others',postPar=None):
|
||||
|
||||
#输入参数
|
||||
# im0s---原始图像列表
|
||||
# model---检测模型，segmodel---分割模型（如若没有用到，则为None）
|
||||
#
|
||||
#输出：两个元素（列表，字符）构成的元组，[im0s[0],im0,det_xywh,iframe],strout
|
||||
# [im0s[0],im0,det_xywh,iframe]中，
|
||||
# im0s[0]--原始图像，im0--AI处理后的图像,iframe--帧号/暂时不需用到。
|
||||
# det_xywh--检测结果，是一个列表。
|
||||
# 其中每一个元素表示一个目标构成如：[ xc,yc,w,h, float(conf_c),float(cls_c) ] ,2023.08.03修改输出格式
|
||||
# #cls_c--类别，如0,1,2,3; xc,yc,w,h--中心点坐标及宽；conf_c--得分, 取值范围在0-1之间
|
||||
# #strout---统计AI处理个环节的时间
|
||||
# Letterbox
|
||||
|
||||
half,device,conf_thres,iou_thres,allowedList = objectPar['half'],objectPar['device'],objectPar['conf_thres'],objectPar['iou_thres'],objectPar['allowedList']
|
||||
|
||||
trtFlag_det,trtFlag_seg,segRegionCnt = objectPar['trtFlag_det'],objectPar['trtFlag_seg'],objectPar['segRegionCnt']
|
||||
if 'ovlap_thres_crossCategory' in objectPar.keys(): ovlap_thres = objectPar['ovlap_thres_crossCategory']
|
||||
else: ovlap_thres = None
|
||||
|
||||
if 'score_byClass' in objectPar.keys(): score_byClass = objectPar['score_byClass']
|
||||
else: score_byClass = None
|
||||
|
||||
time0=time.time()
|
||||
if trtFlag_det:
|
||||
img, padInfos = img_pad(im0s[0], size=(640,640,3)) ;img = [img]
|
||||
else:
|
||||
#print('####line72:',im0s[0][10:12,10:12,2])
|
||||
img = [letterbox(x, 640, auto=True, stride=32)[0] for x in im0s];padInfos=None
|
||||
#print('####line74:',img[0][10:12,10:12,2])
|
||||
# Stack
|
||||
img = np.stack(img, 0)
|
||||
# Convert
|
||||
img = img[:, :, :, ::-1].transpose(0, 3, 1, 2) # BGR to RGB, to bsx3x416x416
|
||||
img = np.ascontiguousarray(img)
|
||||
|
||||
|
||||
img = torch.from_numpy(img).to(device)
|
||||
img = img.half() if half else img.float() # uint8 to fp16/32
|
||||
img /= 255.0
|
||||
time01=time.time()
|
||||
|
||||
|
||||
if segmodel:
|
||||
seg_pred,segstr = segmodel.eval(im0s[0] )
|
||||
segFlag=True
|
||||
else:
|
||||
seg_pred = None;segFlag=False;segstr='Not implemented'
|
||||
|
||||
time1=time.time()
|
||||
if trtFlag_det:
|
||||
pred = yolov5Trtforward(model,img)
|
||||
else:
|
||||
#print('####line96:',img[0,0,10:12,10:12])
|
||||
pred = model(img,augment=False)[0]
|
||||
|
||||
time2=time.time()
|
||||
|
||||
|
||||
|
||||
p_result, timeOut = getDetectionsFromPreds(pred,img,im0s[0],conf_thres=conf_thres,iou_thres=iou_thres,ovlap_thres=ovlap_thres,padInfos=padInfos)
|
||||
#logger.info("获取视频p_result :{}, requestId:{}, names:{}", '+++' * 10, p_result[2],names)
|
||||
|
||||
if score_byClass:
|
||||
p_result[2] = score_filter_byClass(p_result[2],score_byClass)
|
||||
|
||||
|
||||
#print('-'*10,p_result[2])
|
||||
#if mode=='highWay3.0':
|
||||
#if segmodel:
|
||||
if segPar and segPar['mixFunction']['function']:
|
||||
|
||||
mixFunction = segPar['mixFunction']['function'];H,W = im0s[0].shape[0:2]
|
||||
parMix = segPar['mixFunction']['pars'];#print('###line117:',parMix,p_result[2])
|
||||
parMix['imgSize'] = (W,H)
|
||||
#print(' -----------line149: ',p_result[2] ,'\n', seg_pred, parMix ,' sumpSeg:',np.sum(seg_pred))
|
||||
p_result[2] , timeMixPost= mixFunction(p_result[2], seg_pred, pars=parMix )
|
||||
#print(' -----------line112: ',p_result[2] )
|
||||
p_result.append(seg_pred)
|
||||
|
||||
else:
|
||||
timeMixPost=':0 ms'
|
||||
#print('#### line121: segstr:%s timeMixPost:%s timeOut:%s'%( segstr.strip(), timeMixPost,timeOut ))
|
||||
time_info = 'letterbox:%.1f, seg:%.1f , infer:%.1f,%s, seginfo:%s ,timeMixPost:%s '%( (time01-time0)*1000, (time1-time01)*1000 ,(time2-time1)*1000,timeOut , segstr.strip(),timeMixPost )
|
||||
#if mode=='highWay3.0':
|
||||
#print('line159:','-'*10,p_result[2])
|
||||
|
||||
return p_result,time_info
|
||||
def default_mix(predlist,par):
|
||||
return predlist[0],''
|
||||
|
||||
def AI_process_N(im0s,modelList,postProcess):
|
||||
|
||||
#输入参数
|
||||
## im0s---原始图像列表
|
||||
## modelList--所有的模型
|
||||
# postProcess--字典{},包括后处理函数，及其参数
|
||||
#输出参数
|
||||
##ret[0]--检测结果；
|
||||
##ret[1]--时间信息
|
||||
|
||||
#modelList包括模型，每个模型是一个类，里面的eval函数可以输出该模型的推理结果
|
||||
modelRets=[ model.eval(im0s[0]) for model in modelList]
|
||||
|
||||
timeInfos = [ x[1] for x in modelRets]
|
||||
timeInfos=''.join(timeInfos)
|
||||
timeInfos=timeInfos
|
||||
|
||||
#postProcess['function']--后处理函数，输入的就是所有模型输出结果
|
||||
mixFunction =postProcess['function']
|
||||
predsList = [ modelRet[0] for modelRet in modelRets ]
|
||||
H,W = im0s[0].shape[0:2]
|
||||
postProcess['pars']['imgSize'] = (W,H)
|
||||
|
||||
#ret就是混合处理后的结果
|
||||
ret = mixFunction( predsList, postProcess['pars'])
|
||||
|
||||
return ret[0],timeInfos+ret[1]
|
||||
|
||||
def getMaxScoreWords(detRets0):
|
||||
maxScore=-1;maxId=0
|
||||
for i,detRet in enumerate(detRets0):
|
||||
if detRet[4]>maxScore:
|
||||
maxId=i
|
||||
maxScore = detRet[4]
|
||||
return maxId
|
||||
|
||||
|
||||
def AI_process_C(im0s,modelList,postProcess):
|
||||
#函数定制的原因：
|
||||
## 之前模型处理流是
|
||||
## 图片---> 模型1-->result1；图片---> 模型2->result2；[result1,result2]--->后处理函数
|
||||
## 本函数的处理流程是
|
||||
## 图片---> 模型1-->result1；[图片,result1]---> 模型2->result2；[result1,result2]--->后处理函数
|
||||
## 模型2的输入，是有模型1的输出决定的。如模型2是ocr模型，需要将模型1检测出来的船名抠图出来输入到模型2.
|
||||
## 之前的模型流都是模型2是分割模型，输入就是原始图片，与模型1的输出无关。
|
||||
#输入参数
|
||||
## im0s---原始图像列表
|
||||
## modelList--所有的模型
|
||||
# postProcess--字典{},包括后处理函数，及其参数
|
||||
#输出参数
|
||||
##ret[0]--检测结果；
|
||||
##ret[1]--时间信息
|
||||
|
||||
#modelList包括模型，每个模型是一个类，里面的eval函数可以输出该模型的推理结果
|
||||
|
||||
t0=time.time()
|
||||
detRets0 = modelList[0].eval(im0s[0])
|
||||
|
||||
#detRets0=[[12, 46, 1127, 1544, 0.2340087890625, 2.0], [1884, 1248, 2992, 1485, 0.64208984375, 1.0]]
|
||||
detRets0 = detRets0[0]
|
||||
parsIn=postProcess['pars']
|
||||
|
||||
_detRets0_obj = list(filter(lambda x: x[5] in parsIn['objs'], detRets0 ))
|
||||
_detRets0_others = list(filter(lambda x: x[5] not in parsIn['objs'], detRets0 ))
|
||||
_detRets0 = []
|
||||
if postProcess['name']=='channel2':
|
||||
if len(_detRets0_obj)>0:
|
||||
maxId=getMaxScoreWords(_detRets0_obj)
|
||||
_detRets0 = _detRets0_obj[maxId:maxId+1]
|
||||
else: _detRets0 = detRets0
|
||||
|
||||
|
||||
t1=time.time()
|
||||
imagePatches = [ im0s[0][int(x[1]):int(x[3] ) ,int(x[0]):int(x[2])] for x in _detRets0 ]
|
||||
detRets1 = [modelList[1].eval(patch) for patch in imagePatches]
|
||||
print('###line240:',detRets1)
|
||||
if postProcess['name']=='crackMeasurement':
|
||||
detRets1 = [x[0]*255 for x in detRets1]
|
||||
t2=time.time()
|
||||
mixFunction =postProcess['function']
|
||||
crackInfos = [mixFunction(patchMask,par=parsIn) for patchMask in detRets1]
|
||||
|
||||
rets = [ _detRets0[i]+ crackInfos[i] for i in range(len(imagePatches)) ]
|
||||
t3=time.time()
|
||||
outInfos='total:%.1f (det:%.1f %d次segs:%.1f mixProcess:%.1f) '%( (t3-t0)*1000, (t1-t0)*1000, len(detRets1),(t2-t1)*1000, (t3-t2)*1000 )
|
||||
elif postProcess['name']=='channel2':
|
||||
H,W = im0s[0].shape[0:2];parsIn['imgSize'] = (W,H)
|
||||
mixFunction =postProcess['function']
|
||||
_detRets0_others = mixFunction([_detRets0_others], parsIn)
|
||||
ocrInfo='no ocr'
|
||||
if len(_detRets0_obj)>0:
|
||||
res_real = detRets1[0][0]
|
||||
res_real="".join( list(filter(lambda x:(ord(x) >19968 and ord(x)<63865 ) or (ord(x) >47 and ord(x)<58 ),res_real)))
|
||||
|
||||
#detRets1[0][0]="".join( list(filter(lambda x:(ord(x) >19968 and ord(x)<63865 ) or (ord(x) >47 and ord(x)<58 ),detRets1[0][0])))
|
||||
_detRets0_obj[maxId].append(res_real )
|
||||
_detRets0_obj = [_detRets0_obj[maxId]]##只输出有OCR的那个船名结果
|
||||
ocrInfo=detRets1[0][1]
|
||||
print( ' _detRets0_obj:{} _detRets0_others:{} '.format( _detRets0_obj, _detRets0_others ) )
|
||||
rets=_detRets0_obj+_detRets0_others
|
||||
t3=time.time()
|
||||
outInfos='total:%.1f ,where det:%.1f, ocr:%s'%( (t3-t0)*1000, (t1-t0)*1000, ocrInfo)
|
||||
|
||||
#print('###line233:',detRets1,detRets0 )
|
||||
|
||||
return rets,outInfos
|
||||
|
||||
|
||||
|
||||
def AI_process_forest(im0s,model,segmodel,names,label_arraylist,rainbows,half=True,device=' cuda:0',conf_thres=0.25, iou_thres=0.45,allowedList=[0,1,2,3], font={ 'line_thickness':None, 'fontSize':None,'boxLine_thickness':None,'waterLineColor':(0,255,255),'waterLineWidth':3} ,trtFlag_det=False,SecNms=None):
|
||||
#输入参数
|
||||
# im0s---原始图像列表
|
||||
# model---检测模型，segmodel---分割模型（如若没有用到，则为None）
|
||||
#输出：两个元素（列表，字符）构成的元组，[im0s[0],im0,det_xywh,iframe],strout
|
||||
# [im0s[0],im0,det_xywh,iframe]中，
|
||||
# im0s[0]--原始图像，im0--AI处理后的图像,iframe--帧号/暂时不需用到。
|
||||
# det_xywh--检测结果，是一个列表。
|
||||
# 其中每一个元素表示一个目标构成如：[ xc,yc,w,h, float(conf_c),float(cls_c)],#2023.08.03,修改输出格式
|
||||
# #cls_c--类别，如0,1,2,3; xc,yc,w,h--中心点坐标及宽；conf_c--得分, 取值范围在0-1之间
|
||||
# #strout---统计AI处理个环节的时间
|
||||
|
||||
# Letterbox
|
||||
time0=time.time()
|
||||
if trtFlag_det:
|
||||
img, padInfos = img_pad(im0s[0], size=(640,640,3)) ;img = [img]
|
||||
else:
|
||||
img = [letterbox(x, 640, auto=True, stride=32)[0] for x in im0s];padInfos=None
|
||||
#img = [letterbox(x, 640, auto=True, stride=32)[0] for x in im0s]
|
||||
# Stack
|
||||
img = np.stack(img, 0)
|
||||
# Convert
|
||||
img = img[:, :, :, ::-1].transpose(0, 3, 1, 2) # BGR to RGB, to bsx3x416x416
|
||||
img = np.ascontiguousarray(img)
|
||||
|
||||
img = torch.from_numpy(img).to(device)
|
||||
img = img.half() if half else img.float() # uint8 to fp16/32
|
||||
|
||||
img /= 255.0 # 0 - 255 to 0.0 - 1.0
|
||||
if segmodel:
|
||||
seg_pred,segstr = segmodel.eval(im0s[0] )
|
||||
segFlag=True
|
||||
else:
|
||||
seg_pred = None;segFlag=False
|
||||
time1=time.time()
|
||||
pred = yolov5Trtforward(model,img) if trtFlag_det else model(img,augment=False)[0]
|
||||
|
||||
|
||||
time2=time.time()
|
||||
datas = [[''], img, im0s, None,pred,seg_pred,10]
|
||||
|
||||
ObjectPar={ 'object_config':allowedList, 'slopeIndex':[] ,'segmodel':segFlag,'segRegionCnt':0 }
|
||||
p_result,timeOut = post_process_(datas,conf_thres, iou_thres,names,label_arraylist,rainbows,10,ObjectPar=ObjectPar,font=font,padInfos=padInfos,ovlap_thres=SecNms)
|
||||
#print('###line274:',p_result[2])
|
||||
#p_result,timeOut = post_process_(datas,conf_thres, iou_thres,names,label_arraylist,rainbows,10,object_config=allowedList,segmodel=segFlag,font=font,padInfos=padInfos)
|
||||
time_info = 'letterbox:%.1f, infer:%.1f, '%( (time1-time0)*1000,(time2-time1)*1000 )
|
||||
return p_result,time_info+timeOut
|
||||
|
||||
|
||||
def AI_det_track( im0s_in,modelPar,processPar,sort_tracker,segPar=None):
|
||||
im0s,iframe=im0s_in[0],im0s_in[1]
|
||||
model = modelPar['det_Model']
|
||||
segmodel = modelPar['seg_Model']
|
||||
half,device,conf_thres, iou_thres,trtFlag_det = processPar['half'], processPar['device'], processPar['conf_thres'], processPar['iou_thres'],processPar['trtFlag_det']
|
||||
if 'score_byClass' in processPar.keys(): score_byClass = processPar['score_byClass']
|
||||
else: score_byClass = None
|
||||
|
||||
iou2nd = processPar['iou2nd']
|
||||
time0=time.time()
|
||||
|
||||
if trtFlag_det:
|
||||
img, padInfos = img_pad(im0s[0], size=(640,640,3)) ;img = [img]
|
||||
else:
|
||||
img = [letterbox(x, 640, auto=True, stride=32)[0] for x in im0s];padInfos=None
|
||||
img = np.stack(img, 0)
|
||||
# Convert
|
||||
img = img[:, :, :, ::-1].transpose(0, 3, 1, 2) # BGR to RGB, to bsx3x416x416
|
||||
img = np.ascontiguousarray(img)
|
||||
|
||||
img = torch.from_numpy(img).to(device)
|
||||
img = img.half() if half else img.float() # uint8 to fp16/32
|
||||
|
||||
img /= 255.0 # 0 - 255 to 0.0 - 1.0
|
||||
|
||||
seg_pred = None;segFlag=False
|
||||
time1=time.time()
|
||||
pred = yolov5Trtforward(model,img) if trtFlag_det else model(img,augment=False)[0]
|
||||
|
||||
time2=time.time()
|
||||
|
||||
#p_result,timeOut = getDetections(datas,conf_thres, iou_thres,names,label_arraylist,rainbows,10,ObjectPar=ObjectPar,font=font,padInfos=padInfos)
|
||||
p_result, timeOut = getDetectionsFromPreds(pred,img,im0s[0],conf_thres=conf_thres,iou_thres=iou_thres,ovlap_thres=iou2nd,padInfos=padInfos)
|
||||
if score_byClass:
|
||||
p_result[2] = score_filter_byClass(p_result[2],score_byClass)
|
||||
if segmodel:
|
||||
seg_pred,segstr = segmodel.eval(im0s[0] )
|
||||
segFlag=True
|
||||
else:
|
||||
seg_pred = None;segFlag=False;segstr='No segmodel'
|
||||
|
||||
|
||||
if segPar and segPar['mixFunction']['function']:
|
||||
mixFunction = segPar['mixFunction']['function']
|
||||
|
||||
H,W = im0s[0].shape[0:2]
|
||||
parMix = segPar['mixFunction']['pars'];#print('###line117:',parMix,p_result[2])
|
||||
parMix['imgSize'] = (W,H)
|
||||
|
||||
|
||||
p_result[2],timeInfos_post = mixFunction(p_result[2], seg_pred, pars=parMix )
|
||||
timeInfos_seg_post = 'segInfer:%s ,postMixProcess:%s'%( segstr, timeInfos_post )
|
||||
else:
|
||||
timeInfos_seg_post = ' '
|
||||
'''
|
||||
if segmodel:
|
||||
timeS1=time.time()
|
||||
#seg_pred,segstr = segtrtEval(segmodel,im0s[0],par=segPar) if segPar['trtFlag_seg'] else segmodel.eval(im0s[0] )
|
||||
seg_pred,segstr = segmodel.eval(im0s[0] )
|
||||
timeS2=time.time()
|
||||
mixFunction = segPar['mixFunction']['function']
|
||||
|
||||
p_result[2],timeInfos_post = mixFunction(p_result[2], seg_pred, pars=segPar['mixFunction']['pars'] )
|
||||
|
||||
timeInfos_seg_post = 'segInfer:%.1f ,postProcess:%s'%( (timeS2-timeS1)*1000, timeInfos_post )
|
||||
|
||||
else:
|
||||
timeInfos_seg_post = ' '
|
||||
#print('######line341:',seg_pred.shape,np.max(seg_pred),np.min(seg_pred) , len(p_result[2]) )
|
||||
'''
|
||||
time_info = 'letterbox:%.1f, detinfer:%.1f, '%( (time1-time0)*1000,(time2-time1)*1000 )
|
||||
|
||||
if sort_tracker:
|
||||
#在这里增加设置调用追踪器的频率
|
||||
#..................USE TRACK FUNCTION....................
|
||||
#pass an empty array to sort
|
||||
dets_to_sort = np.empty((0,7), dtype=np.float32)
|
||||
|
||||
# NOTE: We send in detected object class too
|
||||
#for detclass,x1,y1,x2,y2,conf in p_result[2]:
|
||||
for x1,y1,x2,y2,conf, detclass in p_result[2]:
|
||||
#print('#######line342:',x1,y1,x2,y2,img.shape,[x1, y1, x2, y2, conf, detclass,iframe])
|
||||
dets_to_sort = np.vstack((dets_to_sort,
|
||||
np.array([x1, y1, x2, y2, conf, detclass,iframe],dtype=np.float32) ))
|
||||
|
||||
# Run SORT
|
||||
tracked_dets = deepcopy(sort_tracker.update(dets_to_sort) )
|
||||
tracks =sort_tracker.getTrackers()
|
||||
p_result.append(tracked_dets) ###index=4
|
||||
p_result.append(tracks) ###index=5
|
||||
|
||||
return p_result,time_info+timeOut+timeInfos_seg_post
|
||||
def AI_det_track_batch(imgarray_list, iframe_list ,modelPar,processPar,sort_tracker,trackPar,segPar=None):
|
||||
'''
|
||||
输入：
|
||||
imgarray_list--图像列表
|
||||
iframe_list -- 帧号列表
|
||||
modelPar--模型参数,字典,modelPar={'det_Model':,'seg_Model':}
|
||||
processPar--字典，存放检测相关参数，'half', 'device', 'conf_thres', 'iou_thres','trtFlag_det'
|
||||
sort_tracker--对象，初始化的跟踪对象。为了保持一致，即使是单帧也要有。
|
||||
trackPar--跟踪参数，关键字包括：det_cnt，windowsize
|
||||
segPar--None,分割模型相关参数。如果用不到，则为None
|
||||
输入：retResults,timeInfos
|
||||
retResults：list
|
||||
retResults[0]--imgarray_list
|
||||
retResults[1]--所有结果用numpy格式，所有的检测结果，包括8类，每列分别是x1, y1, x2, y2, conf, detclass,iframe,trackId
|
||||
retResults[2]--所有结果用list表示,其中每一个元素为一个list，表示每一帧的检测结果，每一个结果是由多个list构成，每个list表示一个框，格式为[ x0 ,y0 ,x1 ,y1 ,conf, cls ,ifrmae,trackId ]，如 retResults[2][j][k]表示第j帧的第k个框。2023.08.03,修改输出格式
|
||||
'''
|
||||
|
||||
det_cnt,windowsize = trackPar['det_cnt'] ,trackPar['windowsize']
|
||||
trackers_dic={}
|
||||
index_list = list(range( 0, len(iframe_list) ,det_cnt ));
|
||||
if len(index_list)>1 and index_list[-1]!= iframe_list[-1]:
|
||||
index_list.append( len(iframe_list) - 1 )
|
||||
|
||||
if len(imgarray_list)==1: #如果是单帧图片，则不用跟踪
|
||||
retResults = []
|
||||
p_result,timeOut = AI_det_track( [ [imgarray_list[0]] ,iframe_list[0] ],modelPar,processPar,None,segPar )
|
||||
##下面4行内容只是为了保持格式一致
|
||||
detArray = np.array(p_result[2])
|
||||
#print('##line371:',detArray)
|
||||
if len(p_result[2])==0:res=[]
|
||||
else:
|
||||
cnt = detArray.shape[0];trackIds=np.zeros((cnt,1));iframes = np.zeros((cnt,1)) + iframe_list[0]
|
||||
|
||||
#detArray = np.hstack( (detArray[:,1:5], detArray[:,5:6] ,detArray[:,0:1],iframes, trackIds ) )
|
||||
detArray = np.hstack( (detArray[:,0:4], detArray[:,4:6] ,iframes, trackIds ) ) ##2023.08.03 修改输入格式
|
||||
res = [[ b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7] ] for b in detArray ]
|
||||
retResults=[imgarray_list,detArray,res ]
|
||||
#print('##line380:',retResults[2])
|
||||
return retResults,timeOut
|
||||
|
||||
else:
|
||||
t0 = time.time()
|
||||
timeInfos_track=''
|
||||
for iframe_index, index_frame in enumerate(index_list):
|
||||
p_result,timeOut = AI_det_track( [ [imgarray_list[index_frame]] ,iframe_list[index_frame] ],modelPar,processPar,sort_tracker,segPar )
|
||||
timeInfos_track='%s:%s'%(timeInfos_track,timeOut)
|
||||
|
||||
for tracker in p_result[5]:
|
||||
trackers_dic[tracker.id]=deepcopy(tracker)
|
||||
t1 = time.time()
|
||||
|
||||
track_det_result = np.empty((0,8))
|
||||
for trackId in trackers_dic.keys():
|
||||
tracker = trackers_dic[trackId]
|
||||
bbox_history = np.array(tracker.bbox_history)
|
||||
if len(bbox_history)<2: continue
|
||||
###把(x0,y0,x1,y1)转换成(xc,yc,w,h)
|
||||
xcs_ycs = (bbox_history[:,0:2] + bbox_history[:,2:4] )/2
|
||||
whs = bbox_history[:,2:4] - bbox_history[:,0:2]
|
||||
bbox_history[:,0:2] = xcs_ycs;bbox_history[:,2:4] = whs;
|
||||
|
||||
arrays_box = bbox_history[:,0:7].transpose();frames=bbox_history[:,6]
|
||||
#frame_min--表示该批次图片的起始帧，如该批次是[1,100],则frame_min=1，[101,200]--frame_min=101
|
||||
#frames[0]--表示该目标出现的起始帧，如[1,11,21,31,41],则frames[0]=1，frames[0]可能会在frame_min之前出现，即一个横跨了多个批次。
|
||||
|
||||
##如果要最好化插值范围，则取内区间[frame_min,则frame_max ]和[frames[0],frames[-1] ]的交集
|
||||
#inter_frame_min = int(max(frame_min, frames[0])); inter_frame_max = int(min( frame_max, frames[-1] )) ##
|
||||
|
||||
##如果要求得到完整的目标轨迹，则插值区间要以目标出现的起始点为准
|
||||
inter_frame_min=int(frames[0]);inter_frame_max=int(frames[-1])
|
||||
new_frames= np.linspace(inter_frame_min,inter_frame_max,inter_frame_max-inter_frame_min+1 )
|
||||
f_linear = interpolate.interp1d(frames,arrays_box); interpolation_x0s = (f_linear(new_frames)).transpose()
|
||||
move_cnt_use =(len(interpolation_x0s)+1)//2*2-1 if len(interpolation_x0s)<windowsize else windowsize
|
||||
for im in range(4):
|
||||
interpolation_x0s[:,im] = moving_average_wang(interpolation_x0s[:,im],move_cnt_use )
|
||||
|
||||
cnt = inter_frame_max-inter_frame_min+1; trackIds = np.zeros((cnt,1)) + trackId
|
||||
interpolation_x0s = np.hstack( (interpolation_x0s, trackIds ) )
|
||||
track_det_result = np.vstack(( track_det_result, interpolation_x0s) )
|
||||
#print('#####line116:',trackId,frame_min,frame_max,'----------',interpolation_x0s.shape,track_det_result.shape ,'-----')
|
||||
|
||||
##将[xc,yc,w,h]转为[x0,y0,x1,y1]
|
||||
x0s = track_det_result[:,0] - track_det_result[:,2]/2 ; x1s = track_det_result[:,0] + track_det_result[:,2]/2
|
||||
y0s = track_det_result[:,1] - track_det_result[:,3]/2 ; y1s = track_det_result[:,1] + track_det_result[:,3]/2
|
||||
track_det_result[:,0] = x0s; track_det_result[:,1] = y0s;
|
||||
track_det_result[:,2] = x1s; track_det_result[:,3] = y1s;
|
||||
detResults=[]
|
||||
for iiframe in iframe_list:
|
||||
boxes_oneFrame = track_det_result[ track_det_result[:,6]==iiframe ]
|
||||
res = [[ b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7] ] for b in boxes_oneFrame ]
|
||||
#[ x0 ,y0 ,x1 ,y1 ,conf,cls,ifrmae,trackId ]
|
||||
#[ifrmae, x0 ,y0 ,x1 ,y1 ,conf,cls,trackId ]
|
||||
detResults.append( res )
|
||||
|
||||
|
||||
retResults=[imgarray_list,track_det_result,detResults ]
|
||||
t2 = time.time()
|
||||
timeInfos = 'detTrack:%.1f TrackPost:%.1f, %s'%(get_ms(t1,t0),get_ms(t2,t1), timeInfos_track )
|
||||
return retResults,timeInfos
|
||||
def AI_det_track_N( im0s_in,modelList,postProcess,sort_tracker):
|
||||
im0s,iframe=im0s_in[0],im0s_in[1]
|
||||
dets = AI_process_N(im0s,modelList,postProcess)
|
||||
p_result=[[],[],dets[0],[] ]
|
||||
if sort_tracker:
|
||||
#在这里增加设置调用追踪器的频率
|
||||
#..................USE TRACK FUNCTION....................
|
||||
#pass an empty array to sort
|
||||
dets_to_sort = np.empty((0,7), dtype=np.float32)
|
||||
|
||||
# NOTE: We send in detected object class too
|
||||
#for detclass,x1,y1,x2,y2,conf in p_result[2]:
|
||||
for x1,y1,x2,y2,conf, detclass in p_result[2]:
|
||||
#print('#######line342:',x1,y1,x2,y2,img.shape,[x1, y1, x2, y2, conf, detclass,iframe])
|
||||
dets_to_sort = np.vstack((dets_to_sort,
|
||||
np.array([x1, y1, x2, y2, conf, detclass,iframe],dtype=np.float32) ))
|
||||
|
||||
# Run SORT
|
||||
tracked_dets = deepcopy(sort_tracker.update(dets_to_sort) )
|
||||
tracks =sort_tracker.getTrackers()
|
||||
p_result.append(tracked_dets) ###index=4
|
||||
p_result.append(tracks) ###index=5
|
||||
|
||||
return p_result,dets[1]
|
||||
def get_tracker_cls(boxes,scId=4,clsId=5):
|
||||
#正常来说一各跟踪链上是一个类别，但是有时目标框检测错误，导致有的跟踪链上有多个类别
|
||||
#为此，根据跟踪链上每一个类别对应的所有框的置信度之和，作为这个跟踪链上目标的类别
|
||||
#输入boxes--跟踪是保留的box_history,[[xc,yc,width,height,score,class,iframe],[...],[...]]
|
||||
## scId=4,score所在的序号; clsId=5;类别所在的序号
|
||||
#输出类别
|
||||
##这个跟踪链上目标的类别
|
||||
ids = list(set(boxes[:,clsId].tolist()))
|
||||
scores = [np.sum( boxes[:,scId] [ boxes[:,clsId]==x ] ) for x in ids]
|
||||
maxScoreId = scores.index(np.max(scores))
|
||||
return int(ids[maxScoreId])
|
||||
|
||||
def AI_det_track_batch_N(imgarray_list, iframe_list ,modelList,postProcess,sort_tracker,trackPar):
|
||||
'''
|
||||
输入：
|
||||
imgarray_list--图像列表
|
||||
iframe_list -- 帧号列表
|
||||
modelPar--模型参数,字典,modelPar={'det_Model':,'seg_Model':}
|
||||
processPar--字典，存放检测相关参数，'half', 'device', 'conf_thres', 'iou_thres','trtFlag_det'
|
||||
sort_tracker--对象，初始化的跟踪对象。为了保持一致，即使是单帧也要有。
|
||||
trackPar--跟踪参数，关键字包括：det_cnt，windowsize
|
||||
segPar--None,分割模型相关参数。如果用不到，则为None
|
||||
输入：retResults,timeInfos
|
||||
retResults：list
|
||||
retResults[0]--imgarray_list
|
||||
retResults[1]--所有结果用numpy格式，所有的检测结果，包括8类，每列分别是x1, y1, x2, y2, conf, detclass,iframe,trackId
|
||||
retResults[2]--所有结果用list表示,其中每一个元素为一个list，表示每一帧的检测结果，每一个结果是由多个list构成，每个list表示一个框，格式为[ x0 ,y0 ,x1 ,y1 ,conf, cls ,ifrmae,trackId ]，如 retResults[2][j][k]表示第j帧的第k个框。2023.08.03,修改输出格式
|
||||
'''
|
||||
|
||||
det_cnt,windowsize = trackPar['det_cnt'] ,trackPar['windowsize']
|
||||
trackers_dic={}
|
||||
index_list = list(range( 0, len(iframe_list) ,det_cnt ));
|
||||
if len(index_list)>1 and index_list[-1]!= iframe_list[-1]:
|
||||
index_list.append( len(iframe_list) - 1 )
|
||||
|
||||
if len(imgarray_list)==1: #如果是单帧图片，则不用跟踪
|
||||
retResults = []
|
||||
p_result,timeOut = AI_det_track_N( [ [imgarray_list[0]] ,iframe_list[0] ],modelList,postProcess,None )
|
||||
##下面4行内容只是为了保持格式一致
|
||||
detArray = np.array(p_result[2])
|
||||
if len(p_result[2])==0:res=[]
|
||||
else:
|
||||
cnt = detArray.shape[0];trackIds=np.zeros((cnt,1));iframes = np.zeros((cnt,1)) + iframe_list[0]
|
||||
|
||||
#detArray = np.hstack( (detArray[:,1:5], detArray[:,5:6] ,detArray[:,0:1],iframes, trackIds ) )
|
||||
detArray = np.hstack( (detArray[:,0:4], detArray[:,4:6] ,iframes, trackIds ) ) ##2023.08.03 修改输入格式
|
||||
res = [[ b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7] ] for b in detArray ]
|
||||
retResults=[imgarray_list,detArray,res ]
|
||||
#print('##line380:',retResults[2])
|
||||
return retResults,timeOut
|
||||
|
||||
else:
|
||||
t0 = time.time()
|
||||
timeInfos_track=''
|
||||
for iframe_index, index_frame in enumerate(index_list):
|
||||
p_result,timeOut = AI_det_track_N( [ [imgarray_list[index_frame]] ,iframe_list[index_frame] ],modelList,postProcess,sort_tracker )
|
||||
timeInfos_track='%s:%s'%(timeInfos_track,timeOut)
|
||||
|
||||
for tracker in p_result[5]:
|
||||
trackers_dic[tracker.id]=deepcopy(tracker)
|
||||
t1 = time.time()
|
||||
|
||||
track_det_result = np.empty((0,8))
|
||||
for trackId in trackers_dic.keys():
|
||||
tracker = trackers_dic[trackId]
|
||||
bbox_history = np.array(tracker.bbox_history).copy()
|
||||
if len(bbox_history)<2: continue
|
||||
###把(x0,y0,x1,y1)转换成(xc,yc,w,h)
|
||||
xcs_ycs = (bbox_history[:,0:2] + bbox_history[:,2:4] )/2
|
||||
whs = bbox_history[:,2:4] - bbox_history[:,0:2]
|
||||
bbox_history[:,0:2] = xcs_ycs;bbox_history[:,2:4] = whs;
|
||||
|
||||
#2023.11.17添加的。目的是修正跟踪链上所有的框的类别一样
|
||||
chainClsId = get_tracker_cls(bbox_history,scId=4,clsId=5)
|
||||
bbox_history[:,5] = chainClsId
|
||||
|
||||
arrays_box = bbox_history[:,0:7].transpose();frames=bbox_history[:,6]
|
||||
#frame_min--表示该批次图片的起始帧，如该批次是[1,100],则frame_min=1，[101,200]--frame_min=101
|
||||
#frames[0]--表示该目标出现的起始帧，如[1,11,21,31,41],则frames[0]=1，frames[0]可能会在frame_min之前出现，即一个横跨了多个批次。
|
||||
|
||||
##如果要最好化插值范围，则取内区间[frame_min,则frame_max ]和[frames[0],frames[-1] ]的交集
|
||||
#inter_frame_min = int(max(frame_min, frames[0])); inter_frame_max = int(min( frame_max, frames[-1] )) ##
|
||||
|
||||
##如果要求得到完整的目标轨迹，则插值区间要以目标出现的起始点为准
|
||||
inter_frame_min=int(frames[0]);inter_frame_max=int(frames[-1])
|
||||
new_frames= np.linspace(inter_frame_min,inter_frame_max,inter_frame_max-inter_frame_min+1 )
|
||||
f_linear = interpolate.interp1d(frames,arrays_box); interpolation_x0s = (f_linear(new_frames)).transpose()
|
||||
move_cnt_use =(len(interpolation_x0s)+1)//2*2-1 if len(interpolation_x0s)<windowsize else windowsize
|
||||
for im in range(4):
|
||||
interpolation_x0s[:,im] = moving_average_wang(interpolation_x0s[:,im],move_cnt_use )
|
||||
|
||||
cnt = inter_frame_max-inter_frame_min+1; trackIds = np.zeros((cnt,1)) + trackId
|
||||
interpolation_x0s = np.hstack( (interpolation_x0s, trackIds ) )
|
||||
track_det_result = np.vstack(( track_det_result, interpolation_x0s) )
|
||||
#print('#####line116:',trackId,'----------',interpolation_x0s.shape,track_det_result.shape,bbox_history ,'-----')
|
||||
|
||||
##将[xc,yc,w,h]转为[x0,y0,x1,y1]
|
||||
x0s = track_det_result[:,0] - track_det_result[:,2]/2 ; x1s = track_det_result[:,0] + track_det_result[:,2]/2
|
||||
y0s = track_det_result[:,1] - track_det_result[:,3]/2 ; y1s = track_det_result[:,1] + track_det_result[:,3]/2
|
||||
track_det_result[:,0] = x0s; track_det_result[:,1] = y0s;
|
||||
track_det_result[:,2] = x1s; track_det_result[:,3] = y1s;
|
||||
detResults=[]
|
||||
for iiframe in iframe_list:
|
||||
boxes_oneFrame = track_det_result[ track_det_result[:,6]==iiframe ]
|
||||
res = [[ b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7] ] for b in boxes_oneFrame ]
|
||||
#[ x0 ,y0 ,x1 ,y1 ,conf,cls,ifrmae,trackId ]
|
||||
#[ifrmae, x0 ,y0 ,x1 ,y1 ,conf,cls,trackId ]
|
||||
detResults.append( res )
|
||||
|
||||
|
||||
retResults=[imgarray_list,track_det_result,detResults ]
|
||||
t2 = time.time()
|
||||
timeInfos = 'detTrack:%.1f TrackPost:%.1f, %s'%(get_ms(t1,t0),get_ms(t2,t1), timeInfos_track )
|
||||
return retResults,timeInfos
|
||||
|
||||
|
||||
|
||||
def ocr_process(pars):
|
||||
|
||||
img_patch,engine,context,converter,AlignCollate_normal,device=pars[0:6]
|
||||
time1 = time.time()
|
||||
img_tensor = AlignCollate_normal([ Image.fromarray(img_patch,'L') ])
|
||||
img_input = img_tensor.to('cuda:0')
|
||||
time2 = time.time()
|
||||
|
||||
preds,trtstr=OcrTrtForward(engine,[img_input],context)
|
||||
time3 = time.time()
|
||||
|
||||
batch_size = preds.size(0)
|
||||
preds_size = torch.IntTensor([preds.size(1)] * batch_size)
|
||||
|
||||
######## filter ignore_char, rebalance
|
||||
preds_prob = F.softmax(preds, dim=2)
|
||||
preds_prob = preds_prob.cpu().detach().numpy()
|
||||
pred_norm = preds_prob.sum(axis=2)
|
||||
preds_prob = preds_prob/np.expand_dims(pred_norm, axis=-1)
|
||||
preds_prob = torch.from_numpy(preds_prob).float().to(device)
|
||||
_, preds_index = preds_prob.max(2)
|
||||
preds_index = preds_index.view(-1)
|
||||
time4 = time.time()
|
||||
preds_str = converter.decode_greedy(preds_index.data.cpu().detach().numpy(), preds_size.data)
|
||||
time5 = time.time()
|
||||
|
||||
info_str= ('pre-process:%.2f TRTforward:%.2f (%s) postProcess:%2.f decoder:%.2f, Total:%.2f , pred:%s'%(get_ms(time2,time1 ),get_ms(time3,time2 ),trtstr, get_ms(time4,time3 ), get_ms(time5,time4 ), get_ms(time5,time1 ), preds_str ) )
|
||||
return preds_str,info_str
|
||||
def main():
|
||||
##预先设置的参数
|
||||
device_='1' ##选定模型，可选 cpu,'0','1'
|
||||
|
||||
##以下参数目前不可改
|
||||
Detweights = "weights/yolov5/class5/best_5classes.pt"
|
||||
seg_nclass = 2
|
||||
Segweights = "weights/BiSeNet/checkpoint.pth"
|
||||
conf_thres,iou_thres,classes= 0.25,0.45,5
|
||||
labelnames = "weights/yolov5/class5/labelnames.json"
|
||||
rainbows = [ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
allowedList=[0,1,2,3]
|
||||
|
||||
|
||||
##加载模型，准备好显示字符
|
||||
device = select_device(device_)
|
||||
names=get_labelnames(labelnames)
|
||||
label_arraylist = get_label_arrays(names,rainbows,outfontsize=40,fontpath="conf/platech.ttf")
|
||||
half = device.type != 'cpu' # half precision only supported on CUDA
|
||||
model = attempt_load(Detweights, map_location=device) # load FP32 model
|
||||
if half: model.half()
|
||||
segmodel = SegModel(nclass=seg_nclass,weights=Segweights,device=device)
|
||||
|
||||
|
||||
##图像测试
|
||||
#url='images/examples/20220624_响水河_12300_1621.jpg'
|
||||
impth = 'images/examples/'
|
||||
outpth = 'images/results/'
|
||||
folders = os.listdir(impth)
|
||||
for i in range(len(folders)):
|
||||
imgpath = os.path.join(impth, folders[i])
|
||||
im0s=[cv2.imread(imgpath)]
|
||||
time00 = time.time()
|
||||
p_result,timeOut = AI_process(im0s,model,segmodel,names,label_arraylist,rainbows,half,device,conf_thres, iou_thres,allowedList,fontSize=1.0)
|
||||
time11 = time.time()
|
||||
image_array = p_result[1]
|
||||
cv2.imwrite( os.path.join( outpth,folders[i] ) ,image_array )
|
||||
#print('----process:%s'%(folders[i]), (time.time() - time11) * 1000)
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
if __name__=="__main__":
|
||||
main()
|
||||
|
|
@ -0,0 +1,45 @@
|
|||
from DMPRUtils.DMPR_process import DMPR_process
|
||||
import tensorrt as trt
|
||||
import sys,os
|
||||
#from DMPRUtils.model.detector import DirectionalPointDetector
|
||||
from DMPRUtils.yolo_net import Model
|
||||
import torch
|
||||
|
||||
class DMPRModel(object):
|
||||
def __init__(self, weights=None,
|
||||
par={'depth_factor':32,'NUM_FEATURE_MAP_CHANNEL':6,'dmpr_thresh':0.3, 'dmprimg_size':640}
|
||||
):
|
||||
|
||||
self.par = par
|
||||
self.device = 'cuda:0'
|
||||
self.half =True
|
||||
|
||||
if weights.endswith('.engine'):
|
||||
self.infer_type ='trt'
|
||||
elif weights.endswith('.pth') or weights.endswith('.pt') :
|
||||
self.infer_type ='pth'
|
||||
else:
|
||||
print('#########ERROR:',weights,': no registered inference type, exit')
|
||||
sys.exit(0)
|
||||
|
||||
if self.infer_type=='trt':
|
||||
logger = trt.Logger(trt.Logger.ERROR)
|
||||
with open(weights, "rb") as f, trt.Runtime(logger) as runtime:
|
||||
self.model=runtime.deserialize_cuda_engine(f.read())# 输入trt本地文件，返回ICudaEngine对象
|
||||
elif self.infer_type=='pth':
|
||||
#self.model = DirectionalPointDetector(3, self.par['depth_factor'], self.par['NUM_FEATURE_MAP_CHANNEL']).to(self.device)
|
||||
confUrl = os.path.join( os.path.dirname(__file__),'DMPRUtils','config','yolov5s.yaml' )
|
||||
self.model = Model(confUrl, ch=3).to(self.device)
|
||||
self.model.load_state_dict(torch.load(weights))
|
||||
print('#######load pt model:%s success '%(weights))
|
||||
self.par['modelType']=self.infer_type
|
||||
print('#########加载模型:',weights,' 类型:',self.infer_type)
|
||||
def eval(self,image):
|
||||
det,timeInfos = DMPR_process(image, self.model, self.device, self.par)
|
||||
det = det.cpu().detach().numpy()
|
||||
return det,timeInfos
|
||||
|
||||
def get_ms(self,t1,t0):
|
||||
return (t1-t0)*1000.0
|
||||
|
||||
|
||||
|
|
@ -0,0 +1,231 @@
|
|||
import math
|
||||
import os
|
||||
import time
|
||||
from collections import namedtuple
|
||||
|
||||
import cv2
|
||||
import numpy as np
|
||||
import torch
|
||||
from torchvision.transforms import ToTensor
|
||||
|
||||
from DMPRUtils.model import DirectionalPointDetector
|
||||
|
||||
from utils.datasets import letterbox
|
||||
from utils.general import clip_coords
|
||||
from utils.torch_utils import select_device
|
||||
#from DMPRUtils.trtUtils import TrtForwardCase
|
||||
#import segutils.trtUtils.segTrtForward as TrtForwardCase
|
||||
from segutils.trtUtils import segTrtForward
|
||||
MarkingPoint = namedtuple('MarkingPoint', ['x', 'y', 'direction', 'shape'])
|
||||
|
||||
|
||||
def plot_points(image, pred_points, line_thickness=3):
|
||||
"""Plot marking points on the image."""
|
||||
if len(pred_points):
|
||||
tl = line_thickness or round(0.002 * (image.shape[0] + image.shape[1]) / 2) + 1 # line/font thickness
|
||||
tf = max(tl - 1, 1) # font thickness
|
||||
for conf, *point in pred_points:
|
||||
p0_x, p0_y = int(point[0]), int(point[1])
|
||||
cos_val = math.cos(point[2])
|
||||
sin_val = math.sin(point[2])
|
||||
p1_x = int(p0_x + 20 * cos_val * tl)
|
||||
p1_y = int(p0_y + 20 * sin_val * tl)
|
||||
p2_x = int(p0_x - 10 * sin_val * tl)
|
||||
p2_y = int(p0_y + 10 * cos_val * tl)
|
||||
p3_x = int(p0_x + 10 * sin_val * tl)
|
||||
p3_y = int(p0_y - 10 * cos_val * tl)
|
||||
|
||||
cv2.line(image, (p0_x, p0_y), (p1_x, p1_y), (0, 0, 255), thickness=tl)
|
||||
cv2.putText(image, str(float(conf)), (p0_x, p0_y), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 0), thickness=tf)
|
||||
if point[3] > 0.5:
|
||||
cv2.line(image, (p0_x, p0_y), (p2_x, p2_y), (0, 0, 255), thickness=tl)
|
||||
else:
|
||||
cv2.line(image, (p2_x, p2_y), (p3_x, p3_y), (0, 0, 255), thickness=tf)
|
||||
|
||||
|
||||
def preprocess_image(image):
|
||||
"""Preprocess numpy image to torch tensor."""
|
||||
if image.shape[0] != 640 or image.shape[1] != 640:
|
||||
image = cv2.resize(image, (640, 640))
|
||||
return torch.unsqueeze(ToTensor()(image), 0)
|
||||
|
||||
def non_maximum_suppression(pred_points):
|
||||
"""Perform non-maxmum suppression on marking points."""
|
||||
t1 = time.time()
|
||||
suppressed = [False] * len(pred_points)
|
||||
for i in range(len(pred_points) - 1):
|
||||
for j in range(i + 1, len(pred_points)):
|
||||
i_x = pred_points[i][1].x
|
||||
i_y = pred_points[i][1].y
|
||||
j_x = pred_points[j][1].x
|
||||
j_y = pred_points[j][1].y
|
||||
# 0.0625 = 1 / 16
|
||||
if abs(j_x - i_x) < 0.0625 and abs(j_y - i_y) < 0.0625:
|
||||
idx = i if pred_points[i][0] < pred_points[j][0] else j
|
||||
suppressed[idx] = True
|
||||
if any(suppressed):
|
||||
unsupres_pred_points = []
|
||||
for i, supres in enumerate(suppressed):
|
||||
if not supres:
|
||||
unsupres_pred_points.append(pred_points[i])
|
||||
return unsupres_pred_points
|
||||
t2 = time.time()
|
||||
print(f'nms: {t2 - t1:.3f}s')
|
||||
return pred_points
|
||||
|
||||
def ms(t2,t1):
|
||||
return ('%.1f '%( (t2-t1)*1000 ) )
|
||||
def get_predicted_points(prediction, thresh):
|
||||
"""Get marking points from one predicted feature map."""
|
||||
t1 = time.time()
|
||||
assert isinstance(prediction, torch.Tensor)
|
||||
|
||||
prediction = prediction.permute(1, 2, 0).contiguous() # prediction (20, 20, 6)
|
||||
height = prediction.shape[0]
|
||||
width = prediction.shape[1]
|
||||
j = torch.arange(prediction.shape[1], device=prediction.device).float().repeat(prediction.shape[0], 1).unsqueeze(dim=2)
|
||||
i = torch.arange(prediction.shape[0], device=prediction.device).float().view(prediction.shape[0], 1).repeat(1,prediction.shape[1]).unsqueeze(dim=2)
|
||||
prediction = torch.cat((prediction, j, i), dim=2).view(-1, 8).contiguous()
|
||||
t2 = time.time()
|
||||
|
||||
# 过滤小于thresh的置信度
|
||||
mask = prediction[..., 0] > thresh
|
||||
t3 = time.time()
|
||||
|
||||
prediction = prediction[mask]
|
||||
t4 = time.time()
|
||||
prediction[..., 2] = (prediction[..., 2] + prediction[..., 6]) / width
|
||||
prediction[..., 3] = (prediction[..., 3] + prediction[..., 7]) / height
|
||||
direction = torch.atan2(prediction[..., 5], prediction[..., 4])
|
||||
prediction = torch.stack((prediction[..., 0], prediction[..., 2], prediction[..., 3], direction, prediction[..., 1]), dim=1)
|
||||
t5 = time.time()
|
||||
timeInfo = 'rerange:%s scoreFilter:%s , getMask:%s stack:%s '%( ms(t2,t1),ms(t3,t2),ms(t4,t3),ms(t5,t4) )
|
||||
#print('-'*20,timeInfo)
|
||||
return prediction,timeInfo
|
||||
|
||||
def get_predicted_points_np(prediction, thresh):
|
||||
"""Get marking points from one predicted feature map."""
|
||||
t1 = time.time()
|
||||
prediction = prediction.permute(1, 2, 0).contiguous() # prediction (20, 20, 6)
|
||||
t1_1 = time.time()
|
||||
prediction = prediction.cpu().detach().numpy()
|
||||
t1_2 = time.time()
|
||||
height,width = prediction.shape[0:2]
|
||||
i,j = np.mgrid[0:height, 0:width]
|
||||
i = np.expand_dims(i,axis=2);j = np.expand_dims(j,axis=2)
|
||||
#print('##line112:',i.shape,j.shape,prediction.shape)
|
||||
prediction = np.concatenate( (prediction,i,j),axis=2 )
|
||||
prediction = prediction.reshape(-1,8)
|
||||
t2 = time.time()
|
||||
mask = prediction[..., 0] > thresh
|
||||
t3 = time.time()
|
||||
|
||||
prediction = prediction[mask]
|
||||
t4 = time.time()
|
||||
prediction[..., 2] = (prediction[..., 2] + prediction[..., 6]) / width
|
||||
prediction[..., 3] = (prediction[..., 3] + prediction[..., 7]) / height
|
||||
direction = np.arctan(prediction[..., 5:6], prediction[..., 4:5])
|
||||
#print('-'*20,prediction.shape,direction.shape)
|
||||
prediction = np.hstack((prediction[:, 0:1], prediction[:, 2:3], prediction[:, 3:4], direction, prediction[:, 1:2]))
|
||||
#print('-line126:','-'*20,type(prediction),prediction.shape)
|
||||
|
||||
t5 = time.time()
|
||||
|
||||
|
||||
timeInfo = 'permute:%s Tocpu:%s rerange:%s scoreFilter:%s , getMask:%s stack:%s '%( ms(t1_1,t1) , ms(t1_2,t1_1),ms(t2,t1_2),ms(t3,t2),ms(t4,t3),ms(t5,t4) )
|
||||
print('-'*20,timeInfo,prediction.shape)
|
||||
return prediction
|
||||
|
||||
|
||||
def detect_marking_points(detector, image, thresh, device,modelType='pth'):
|
||||
"""Given image read from opencv, return detected marking points."""
|
||||
t1 = time.time()
|
||||
|
||||
image_preprocess = preprocess_image(image).to(device)
|
||||
if modelType=='pth':
|
||||
|
||||
prediction = detector(image_preprocess)
|
||||
#print(prediction)
|
||||
elif modelType=='trt':
|
||||
a=0
|
||||
prediction = segTrtForward(detector,[image_preprocess ])
|
||||
#print(prediction)
|
||||
|
||||
|
||||
torch.cuda.synchronize(device)
|
||||
t2 = time.time()
|
||||
|
||||
rets,timeInfo = get_predicted_points(prediction[0], thresh)
|
||||
string_t2 = ' infer:%s postprocess:%s'%(ms(t2,t1),timeInfo)
|
||||
|
||||
return rets
|
||||
|
||||
def scale_coords2(img1_shape, coords, img0_shape, ratio_pad=None):
|
||||
# Rescale coords (xy) from img1_shape to img0_shape
|
||||
if ratio_pad is None: # calculate from img0_shape
|
||||
gain = min(img1_shape[0] / img0_shape[0], img1_shape[1] / img0_shape[1]) # gain = old / new
|
||||
pad = (img1_shape[1] - img0_shape[1] * gain) / 2, (img1_shape[0] - img0_shape[0] * gain) / 2 # wh padding
|
||||
else:
|
||||
gain = ratio_pad[0][0]
|
||||
pad = ratio_pad[1]
|
||||
# 百分比x，y转换为实际x，y
|
||||
height, width = img1_shape
|
||||
if isinstance(coords, torch.Tensor):
|
||||
coords[:, 0] = torch.round(width * coords[:, 0] - 0.5)
|
||||
coords[:, 1] = torch.round(height * coords[:, 1] - 0.5)
|
||||
else:
|
||||
coords[:, 0] = (width * coords[:, 0] + 0.5).astype(np.int32)
|
||||
coords[:, 1] = (height * coords[:, 1] + 0.5).astype(np.int32)
|
||||
|
||||
coords[:, 0] -= pad[0] # x padding
|
||||
coords[:, 1] -= pad[1] # y padding
|
||||
coords[:, :3] /= gain
|
||||
#恢复成原始图片尺寸
|
||||
if isinstance(coords, torch.Tensor):
|
||||
coords[:, 0].clamp_(0, img0_shape[1])
|
||||
coords[:, 1].clamp_(0, img0_shape[0])
|
||||
else:
|
||||
coords[:, 0] = np.clip( coords[:, 0], 0,img0_shape[1] )
|
||||
coords[:, 1] = np.clip( coords[:, 1], 0,img0_shape[0] )
|
||||
|
||||
return coords
|
||||
|
||||
|
||||
def DMPR_process(img0, model, device, DMPRmodelPar):
|
||||
t0 = time.time()
|
||||
height, width, _ = img0.shape
|
||||
|
||||
img, ratio, (dw, dh) = letterbox(img0, DMPRmodelPar['dmprimg_size'], auto=False)
|
||||
t1 = time.time()
|
||||
#print('###line188:', height, width, img.shape)
|
||||
det = detect_marking_points(model, img, DMPRmodelPar['dmpr_thresh'], device,modelType=DMPRmodelPar['modelType'])
|
||||
t2 = time.time()
|
||||
if len(det):
|
||||
det[:, 1:3] = scale_coords2(img.shape[:2], det[:, 1:3], img0.shape)
|
||||
|
||||
|
||||
t3 = time.time()
|
||||
timeInfos = 'dmpr:%1.f (lettbox:%.1f dectect:%.1f scaleBack:%.1f) '%( (t3-t0)*1000,(t1-t0)*1000,(t2-t1)*1000,(t3-t2)*1000, )
|
||||
|
||||
return det,timeInfos
|
||||
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
impath = r'I:\zjc\weiting1\Images'
|
||||
file = 'DJI_0001_8.jpg'
|
||||
imgpath = os.path.join(impath, file)
|
||||
img0 = cv2.imread(imgpath)
|
||||
|
||||
device_ = '0'
|
||||
device = select_device(device_)
|
||||
args = config.get_parser_for_inference().parse_args()
|
||||
model = DirectionalPointDetector(3, args.depth_factor, config.NUM_FEATURE_MAP_CHANNEL).to(device)
|
||||
weights = r"E:\pycharmProject\DMPR-PS\weights\dp_detector_499.pth"
|
||||
model.load_state_dict(torch.load(weights))
|
||||
|
||||
det = DMPR_process(img0, model, device, args)
|
||||
|
||||
plot_points(img0, det)
|
||||
|
||||
cv2.imwrite(file, img0, [int(cv2.IMWRITE_JPEG_QUALITY), 100])
|
||||
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
|
|
@ -0,0 +1,48 @@
|
|||
# parameters
|
||||
nc: 80 # number of classes
|
||||
depth_multiple: 1.0 # model depth multiple
|
||||
width_multiple: 1.0 # layer channel multiple
|
||||
|
||||
# anchors
|
||||
anchors:
|
||||
- [10,13, 16,30, 33,23] # P3/8
|
||||
- [30,61, 62,45, 59,119] # P4/16
|
||||
- [116,90, 156,198, 373,326] # P5/32
|
||||
|
||||
# YOLOv5 backbone
|
||||
backbone:
|
||||
# [from, number, module, args]
|
||||
[[-1, 1, Focus, [64, 3]], # 0-P1/2
|
||||
[-1, 1, Conv, [128, 3, 2]], # 1-P2/4
|
||||
[-1, 3, C3, [128]],
|
||||
[-1, 1, Conv, [256, 3, 2]], # 3-P3/8
|
||||
[-1, 9, C3, [256]],
|
||||
[-1, 1, Conv, [512, 3, 2]], # 5-P4/16
|
||||
[-1, 9, C3, [512]],
|
||||
[-1, 1, Conv, [1024, 3, 2]], # 7-P5/32
|
||||
[-1, 1, SPP, [1024, [5, 9, 13]]],
|
||||
[-1, 3, C3, [1024, False]], # 9
|
||||
]
|
||||
|
||||
# YOLOv5 head
|
||||
head:
|
||||
[[-1, 1, Conv, [512, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 6], 1, Concat, [1]], # cat backbone P4
|
||||
[-1, 3, C3, [512, False]], # 13
|
||||
|
||||
[-1, 1, Conv, [256, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 4], 1, Concat, [1]], # cat backbone P3
|
||||
[-1, 3, C3, [256, False]], # 17 (P3/8-small)
|
||||
|
||||
[-1, 1, Conv, [256, 3, 2]],
|
||||
[[-1, 14], 1, Concat, [1]], # cat head P4
|
||||
[-1, 3, C3, [512, False]], # 20 (P4/16-medium)
|
||||
|
||||
[-1, 1, Conv, [512, 3, 2]],
|
||||
[[-1, 10], 1, Concat, [1]], # cat head P5
|
||||
[-1, 3, C3, [1024, False]], # 23 (P5/32-large)
|
||||
|
||||
[[17, 20, 23], 1, Detect, [nc, anchors]], # Detect(P3, P4, P5)
|
||||
]
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
# parameters
|
||||
nc: 80 # number of classes
|
||||
depth_multiple: 0.67 # model depth multiple
|
||||
width_multiple: 0.75 # layer channel multiple
|
||||
|
||||
# anchors
|
||||
anchors:
|
||||
- [10,13, 16,30, 33,23] # P3/8
|
||||
- [30,61, 62,45, 59,119] # P4/16
|
||||
- [116,90, 156,198, 373,326] # P5/32
|
||||
|
||||
# YOLOv5 backbone
|
||||
backbone:
|
||||
# [from, number, module, args]
|
||||
[[-1, 1, Focus, [64, 3]], # 0-P1/2
|
||||
[-1, 1, Conv, [128, 3, 2]], # 1-P2/4
|
||||
[-1, 3, C3, [128]],
|
||||
[-1, 1, Conv, [256, 3, 2]], # 3-P3/8
|
||||
[-1, 9, C3, [256]],
|
||||
[-1, 1, Conv, [512, 3, 2]], # 5-P4/16
|
||||
[-1, 9, C3, [512]],
|
||||
[-1, 1, Conv, [1024, 3, 2]], # 7-P5/32
|
||||
[-1, 1, SPP, [1024, [5, 9, 13]]],
|
||||
[-1, 3, C3, [1024, False]], # 9
|
||||
]
|
||||
|
||||
# YOLOv5 head
|
||||
head:
|
||||
[[-1, 1, Conv, [512, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 6], 1, Concat, [1]], # cat backbone P4
|
||||
[-1, 3, C3, [512, False]], # 13
|
||||
|
||||
[-1, 1, Conv, [256, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 4], 1, Concat, [1]], # cat backbone P3
|
||||
[-1, 3, C3, [256, False]], # 17 (P3/8-small)
|
||||
|
||||
[-1, 1, Conv, [256, 3, 2]],
|
||||
[[-1, 14], 1, Concat, [1]], # cat head P4
|
||||
[-1, 3, C3, [512, False]], # 20 (P4/16-medium)
|
||||
|
||||
[-1, 1, Conv, [512, 3, 2]],
|
||||
[[-1, 10], 1, Concat, [1]], # cat head P5
|
||||
[-1, 3, C3, [1024, False]], # 23 (P5/32-large)
|
||||
|
||||
[[17, 20, 23], 1, Detect, [nc, anchors]], # Detect(P3, P4, P5)
|
||||
]
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
# parameters
|
||||
nc: 80 # number of classes
|
||||
depth_multiple: 0.33 # model depth multiple
|
||||
width_multiple: 0.50 # layer channel multiple
|
||||
|
||||
# anchors
|
||||
anchors:
|
||||
- [10,13, 16,30, 33,23] # P3/8
|
||||
- [30,61, 62,45, 59,119] # P4/16
|
||||
- [116,90, 156,198, 373,326] # P5/32
|
||||
|
||||
# YOLOv5 backbone
|
||||
backbone:
|
||||
# [from, number, module, args]
|
||||
[[-1, 1, Focus, [64, 3]], # 0-P1/2
|
||||
[-1, 1, Conv, [128, 3, 2]], # 1-P2/4
|
||||
[-1, 3, C3, [128]],
|
||||
[-1, 1, Conv, [256, 3, 2]], # 3-P3/8
|
||||
[-1, 9, C3, [256]],
|
||||
[-1, 1, Conv, [512, 3, 2]], # 5-P4/16
|
||||
[-1, 9, C3, [512]],
|
||||
[-1, 1, Conv, [1024, 3, 2]], # 7-P5/32
|
||||
[-1, 1, SPP, [1024, [5, 9, 13]]],
|
||||
[-1, 3, C3, [1024, False]], # 9
|
||||
]
|
||||
|
||||
# YOLOv5 head
|
||||
head:
|
||||
[[-1, 1, Conv, [512, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 6], 1, Concat, [1]], # cat backbone P4
|
||||
[-1, 3, C3, [512, False]], # 13
|
||||
|
||||
[-1, 1, Conv, [256, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 4], 1, Concat, [1]], # cat backbone P3
|
||||
[-1, 3, C3, [256, False]], # 17 (P3/8-small)
|
||||
|
||||
[-1, 1, Conv, [256, 3, 2]],
|
||||
[[-1, 14], 1, Concat, [1]], # cat head P4
|
||||
[-1, 3, C3, [512, False]], # 20 (P4/16-medium)
|
||||
|
||||
[-1, 1, Conv, [512, 3, 2]],
|
||||
[[-1, 10], 1, Concat, [1]], # cat head P5
|
||||
[-1, 3, C3, [1024, False]], # 23 (P5/32-large)
|
||||
|
||||
[[17, 20, 23], 1, Detect, [nc, anchors]], # Detect(P3, P4, P5)
|
||||
]
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
# parameters
|
||||
nc: 80 # number of classes
|
||||
depth_multiple: 1.33 # model depth multiple
|
||||
width_multiple: 1.25 # layer channel multiple
|
||||
|
||||
# anchors
|
||||
anchors:
|
||||
- [10,13, 16,30, 33,23] # P3/8
|
||||
- [30,61, 62,45, 59,119] # P4/16
|
||||
- [116,90, 156,198, 373,326] # P5/32
|
||||
|
||||
# YOLOv5 backbone
|
||||
backbone:
|
||||
# [from, number, module, args]
|
||||
[[-1, 1, Focus, [64, 3]], # 0-P1/2
|
||||
[-1, 1, Conv, [128, 3, 2]], # 1-P2/4
|
||||
[-1, 3, C3, [128]],
|
||||
[-1, 1, Conv, [256, 3, 2]], # 3-P3/8
|
||||
[-1, 9, C3, [256]],
|
||||
[-1, 1, Conv, [512, 3, 2]], # 5-P4/16
|
||||
[-1, 9, C3, [512]],
|
||||
[-1, 1, Conv, [1024, 3, 2]], # 7-P5/32
|
||||
[-1, 1, SPP, [1024, [5, 9, 13]]],
|
||||
[-1, 3, C3, [1024, False]], # 9
|
||||
]
|
||||
|
||||
# YOLOv5 head
|
||||
head:
|
||||
[[-1, 1, Conv, [512, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 6], 1, Concat, [1]], # cat backbone P4
|
||||
[-1, 3, C3, [512, False]], # 13
|
||||
|
||||
[-1, 1, Conv, [256, 1, 1]],
|
||||
[-1, 1, nn.Upsample, [None, 2, 'nearest']],
|
||||
[[-1, 4], 1, Concat, [1]], # cat backbone P3
|
||||
[-1, 3, C3, [256, False]], # 17 (P3/8-small)
|
||||
|
||||
[-1, 1, Conv, [256, 3, 2]],
|
||||
[[-1, 14], 1, Concat, [1]], # cat head P4
|
||||
[-1, 3, C3, [512, False]], # 20 (P4/16-medium)
|
||||
|
||||
[-1, 1, Conv, [512, 3, 2]],
|
||||
[[-1, 10], 1, Concat, [1]], # cat head P5
|
||||
[-1, 3, C3, [1024, False]], # 23 (P5/32-large)
|
||||
|
||||
[[17, 20, 23], 1, Detect, [nc, anchors]], # Detect(P3, P4, P5)
|
||||
]
|
||||
|
|
@ -0,0 +1,149 @@
|
|||
import math
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
import time
|
||||
|
||||
|
||||
def dmpr_yolo( yolo_det, dmpr_det,pars):
|
||||
#if len(yolo_det)==0 or len(dmpr_det)==0:
|
||||
|
||||
#print('line11:\n',yolo_det, dmpr_det,pars)
|
||||
time1=time.time()
|
||||
if len(yolo_det)==0:
|
||||
return yolo_det,' No yolo detections'
|
||||
|
||||
img_shape = (pars['imgSize'][1],pars['imgSize'][0])
|
||||
cls = pars['carCls']; scaleRatio = pars['scaleRatio']
|
||||
illParkCls = pars['illCls'];border = pars['border']
|
||||
|
||||
yolo_det = np.array(yolo_det)
|
||||
yolo_det_0 = yolo_det.copy()
|
||||
|
||||
|
||||
|
||||
|
||||
#print('-'*10,'line17：',yolo_det_0)
|
||||
|
||||
# 过滤在图像边界的box(防止出现类似一小半车辆的情况)
|
||||
|
||||
x_c = (yolo_det[:, 0] + yolo_det[:, 2]) / 2
|
||||
y_c = (yolo_det[:, 1] + yolo_det[:, 3]) / 2
|
||||
tmp = (x_c >= border) & (x_c <= (img_shape[1] - border)) & (y_c >= border) & (y_c <= (img_shape[0] - border))
|
||||
yolo_det = yolo_det[tmp]
|
||||
|
||||
|
||||
# 创建yolo_det_clone内容为x1, y1, x2, y2, conf, cls, unlabel (unlabel代表该类是否需要忽略，0：不忽略 其他：忽略)
|
||||
yolo_det_clone = yolo_det.copy()
|
||||
tmp_0_tensor = np.zeros([len(yolo_det), 1])
|
||||
yolo_det_clone = np.concatenate([yolo_det_clone, tmp_0_tensor], axis=1)
|
||||
|
||||
# cls为需要计算的类别
|
||||
yolo_det = yolo_det[yolo_det[:, -1] == cls]
|
||||
|
||||
# new_yolo_det为膨胀后数据，内容为x1, y1, x2, y2, flag (flag代表膨胀后车位内是否包含角点 且 与角点方向差值小于90度, 其值为第一个满足条件的角点索引)
|
||||
new_yolo_det = np.zeros([len(yolo_det), 7])
|
||||
|
||||
# yolo框膨胀，长的边两边各膨胀0.4倍总长，短的边两边各膨胀0.2倍总长
|
||||
x_length = yolo_det[:, 2] - yolo_det[:, 0] #x2-x1
|
||||
y_length = yolo_det[:, 3] - yolo_det[:, 1] #y2-y1
|
||||
|
||||
# x, y哪个方向差值大哪个方向膨胀的多
|
||||
x_dilate_coefficient = ((x_length > y_length) + 1)*scaleRatio
|
||||
y_dilate_coefficient = ((~(x_length > y_length)) + 1)*scaleRatio
|
||||
|
||||
# 原始框中心点x_c, y_c
|
||||
new_yolo_det[:, 5] = (yolo_det[:, 0] + yolo_det[:, 2]) / 2
|
||||
new_yolo_det[:, 6] = (yolo_det[:, 1] + yolo_det[:, 3]) / 2
|
||||
|
||||
# 膨胀
|
||||
new_yolo_det[:, 0] = np.round(yolo_det[:, 0] - x_dilate_coefficient * x_length).clip(0, img_shape[1]) #x1 膨胀
|
||||
new_yolo_det[:, 1] = np.round(yolo_det[:, 1] - y_dilate_coefficient * y_length).clip(0, img_shape[0]) #y1 膨胀
|
||||
new_yolo_det[:, 2] = np.round(yolo_det[:, 2] + x_dilate_coefficient * x_length).clip(0, img_shape[1]) #x2 膨胀
|
||||
new_yolo_det[:, 3] = np.round(yolo_det[:, 3] + y_dilate_coefficient * y_length).clip(0, img_shape[0]) #y2 膨胀
|
||||
|
||||
m, n = new_yolo_det.size, dmpr_det.size
|
||||
|
||||
if not m or not n:
|
||||
#print('##line47 original yolo_det_clone:',yolo_det_clone)
|
||||
yolo_det_clone[np.logical_and( yolo_det_clone[:,-1]==0,yolo_det_clone[:,-2]==cls),-2] = illParkCls
|
||||
|
||||
#yolo_det_clone[yolo_det_clone[:, -1] == 0 & yolo_det_clone[:, -2==cls] , -2] = illParkCls
|
||||
return yolo_det_clone[:,0:6], ' no cars or T/L corners'
|
||||
|
||||
new_yolo = new_yolo_det[:, np.newaxis, :].repeat(dmpr_det.shape[0], 1) # 扩展为 (m , n, 5)
|
||||
dmpr_det = dmpr_det[np.newaxis, ...].repeat(new_yolo_det.shape[0], 0)
|
||||
yolo_dmpr = np.concatenate((new_yolo, dmpr_det), axis=2) # (m, n, 10)
|
||||
|
||||
x_p, y_p = yolo_dmpr[..., 8], yolo_dmpr[..., 9]
|
||||
x1, y1, x2, y2 = yolo_dmpr[..., 0], yolo_dmpr[..., 1], yolo_dmpr[..., 2], yolo_dmpr[..., 3]
|
||||
x_c, y_c = yolo_dmpr[..., 5], yolo_dmpr[..., 6]
|
||||
|
||||
direction1 = np.arctan2(y_c - y_p, x_c - x_p) / math.pi * 180
|
||||
direction2 = yolo_dmpr[..., 10] / math.pi * 180
|
||||
direction3 = direction2 + 90 # L形角点另外一个方向
|
||||
direction3[direction3 > 180] -= 360
|
||||
ang_diff = direction1 - direction2
|
||||
ang_diff2 = direction1 - direction3
|
||||
|
||||
# 判断膨胀后yolo框包含角点关系 && 包含角点的时候计算水平框中心点与角点的角度关系
|
||||
# direction ∈ （-180， 180） 若角差大于180，需算补角
|
||||
# T形角点比较一个方向，L形角点比较两个方向
|
||||
mask = (x_p >= x1) & (x_p <= x2) & (y_p >= y1) & (y_p <= y2) & \
|
||||
(((yolo_dmpr[..., 11] <= 0.5) & # T形角点情况
|
||||
(((ang_diff >= -90) & (ang_diff <= 90)) | ((ang_diff > 180) & ((360 - ang_diff) <= 90)) |
|
||||
(((ang_diff) < -180) & ((360 + ang_diff) <= 90)))) |
|
||||
((yolo_dmpr[..., 11] > 0.5) & # L形角点情况
|
||||
(((ang_diff >= -90) & (ang_diff <= 90)) | ((ang_diff > 180) & ((360 - ang_diff) <= 90)) |
|
||||
(((ang_diff) < -180) & ((360 + ang_diff) <= 90))) &
|
||||
(((ang_diff2 >= -90) & (ang_diff2 <= 90)) | ((ang_diff2 > 180) & ((360 - ang_diff2) <= 90)) |
|
||||
(((ang_diff2) < -180) & ((360 + ang_diff2) <= 90)))))
|
||||
|
||||
res = np.sum(mask, axis=1)
|
||||
|
||||
yolo_det_clone[yolo_det_clone[:, -2] == cls, -1] = res
|
||||
#print('##line69 original yolo_det_clone:',yolo_det_clone)
|
||||
#yolo_det_clone[yolo_det_clone[:, -1] == 0, -2] = illParkCls
|
||||
|
||||
#print('-'*20,'--line78',yolo_det_clone)
|
||||
yolo_det_clone[ np.logical_and( yolo_det_clone[:,-1]==0,yolo_det_clone[:,-2]==cls) ,-2 ] = illParkCls
|
||||
#print('-'*20,'--line80:',yolo_det_clone)
|
||||
yolo_det_clone = yolo_det_clone[:,0:6]
|
||||
time2=time.time()
|
||||
|
||||
return np.array(yolo_det_clone), 'dmpr_yolo:%.1f'%( (time2-time1)*1000 )
|
||||
def stdc_yolo(stdc_det, yolo_det,pars):
|
||||
|
||||
is_car = yolo_det[:, -1] == pars['carCls'] # 获取最后一列，判断是否等于0
|
||||
car = yolo_det[is_car] # 筛选出最后一列等于0的行
|
||||
no_car = yolo_det[~is_car] # 筛选出最后一列不等于0的行
|
||||
|
||||
im = np.uint8(stdc_det)
|
||||
x_c = ((car[:, 0] + car[:, 2]) // 2).astype(int)
|
||||
y_c = ((car[:, 1] + car[:, 3]) // 2).astype(int)
|
||||
car_filted = car[im[y_c, x_c] == 0]
|
||||
#yolo_filted = yolo_det
|
||||
|
||||
yolo_filted = np.concatenate((car_filted, no_car), axis=0)
|
||||
return yolo_filted
|
||||
|
||||
def dmpr_yolo_stdc(predsList,pars):
|
||||
if len(predsList)==2:
|
||||
yolo_det, dmpr_det = predsList[0:2]
|
||||
else:
|
||||
yolo_det, dmpr_det,stdc_det = predsList[0:3]
|
||||
if len(yolo_det)==0:
|
||||
return yolo_det,' No yolo detections'
|
||||
if isinstance(yolo_det,list):
|
||||
yolo_det = np.array(yolo_det)
|
||||
if len(predsList)>2:
|
||||
yolo_det = stdc_yolo(stdc_det, yolo_det,pars)
|
||||
|
||||
rets = dmpr_yolo(yolo_det, dmpr_det,pars)
|
||||
for i,ret in enumerate(rets[0]):
|
||||
#print(ret,'\n ',rets,pars['classReindex'])
|
||||
ret[5] = pars['classReindex'][ret[5]]
|
||||
#rets[i][5] = pars['classReindex'][ret[5]]
|
||||
|
||||
|
||||
return rets
|
||||
|
|
@ -0,0 +1,142 @@
|
|||
import math
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
import time
|
||||
from loguru import logger
|
||||
|
||||
|
||||
def dmpr_yolo( yolo_det, dmpr_det,pars):
|
||||
#if len(yolo_det)==0 or len(dmpr_det)==0:
|
||||
|
||||
#print('line11:\n',yolo_det, dmpr_det,pars)
|
||||
time1=time.time()
|
||||
if len(yolo_det)==0:
|
||||
return yolo_det,' No yolo detections'
|
||||
|
||||
img_shape = (pars['imgSize'][1],pars['imgSize'][0])
|
||||
cls = pars['carCls']; scaleRatio = pars['scaleRatio']
|
||||
illParkCls = pars['illCls'];border = pars['border']
|
||||
|
||||
yolo_det = np.array(yolo_det)
|
||||
yolo_det_0 = yolo_det.copy()
|
||||
|
||||
|
||||
|
||||
|
||||
#print('-'*10,'line17：',yolo_det_0)
|
||||
|
||||
# 过滤在图像边界的box(防止出现类似一小半车辆的情况)
|
||||
|
||||
x_c = (yolo_det[:, 0] + yolo_det[:, 2]) / 2
|
||||
y_c = (yolo_det[:, 1] + yolo_det[:, 3]) / 2
|
||||
tmp = (x_c >= border) & (x_c <= (img_shape[1] - border)) & (y_c >= border) & (y_c <= (img_shape[0] - border))
|
||||
yolo_det = yolo_det[tmp]
|
||||
|
||||
|
||||
# 创建yolo_det_clone内容为x1, y1, x2, y2, conf, cls, unlabel (unlabel代表该类是否需要忽略，0：不忽略 其他：忽略)
|
||||
logger.info("标记 :{}, yolo_det:{}, dmpr_det:{}", '+++'* 10, yolo_det, dmpr_det)
|
||||
yolo_det_clone = yolo_det.copy()
|
||||
tmp_0_tensor = np.zeros([len(yolo_det), 1])
|
||||
yolo_det_clone = np.concatenate([yolo_det_clone, tmp_0_tensor], axis=1)
|
||||
|
||||
# cls为需要计算的类别
|
||||
yolo_det = yolo_det[yolo_det[:, -1] == cls]
|
||||
|
||||
# new_yolo_det为膨胀后数据，内容为x1, y1, x2, y2, flag (flag代表膨胀后车位内是否包含角点 且 与角点方向差值小于90度, 其值为第一个满足条件的角点索引)
|
||||
new_yolo_det = np.zeros([len(yolo_det), 7])
|
||||
|
||||
# yolo框膨胀，长的边两边各膨胀0.4倍总长，短的边两边各膨胀0.2倍总长
|
||||
x_length = yolo_det[:, 2] - yolo_det[:, 0] #x2-x1
|
||||
y_length = yolo_det[:, 3] - yolo_det[:, 1] #y2-y1
|
||||
|
||||
# x, y哪个方向差值大哪个方向膨胀的多
|
||||
x_dilate_coefficient = ((x_length > y_length) + 1)*scaleRatio
|
||||
y_dilate_coefficient = ((~(x_length > y_length)) + 1)*scaleRatio
|
||||
|
||||
|
||||
logger.info("膨胀前标记 :{}, yolo_det:{}", '+++'*10, new_yolo_det)
|
||||
# 原始框中心点x_c, y_c
|
||||
new_yolo_det[:, 5] = (yolo_det[:, 0] + yolo_det[:, 2]) / 2
|
||||
new_yolo_det[:, 6] = (yolo_det[:, 1] + yolo_det[:, 3]) / 2
|
||||
|
||||
# 膨胀
|
||||
new_yolo_det[:, 0] = np.round(yolo_det[:, 0] - x_dilate_coefficient * x_length).clip(0, img_shape[1]) #x1 膨胀
|
||||
new_yolo_det[:, 1] = np.round(yolo_det[:, 1] - y_dilate_coefficient * y_length).clip(0, img_shape[0]) #y1 膨胀
|
||||
new_yolo_det[:, 2] = np.round(yolo_det[:, 2] + x_dilate_coefficient * x_length).clip(0, img_shape[1]) #x2 膨胀
|
||||
new_yolo_det[:, 3] = np.round(yolo_det[:, 3] + y_dilate_coefficient * y_length).clip(0, img_shape[0]) #y2 膨胀
|
||||
|
||||
m, n = new_yolo_det.size, dmpr_det.size
|
||||
|
||||
logger.info("膨胀后标记 :{}, yolo_det:{}， m:{}, n:{}", '+++'*10, new_yolo_det,m,n)
|
||||
if not m or not n:
|
||||
#print('##line47 original yolo_det_clone:',yolo_det_clone)
|
||||
yolo_det_clone[np.logical_and( yolo_det_clone[:,-1]==0,yolo_det_clone[:,-2]==cls),-2] = illParkCls
|
||||
|
||||
#yolo_det_clone[yolo_det_clone[:, -1] == 0 & yolo_det_clone[:, -2==cls] , -2] = illParkCls
|
||||
return yolo_det_clone[:,0:6], ' no cars or T/L corners'
|
||||
|
||||
new_yolo = new_yolo_det[:, np.newaxis, :].repeat(dmpr_det.shape[0], 1) # 扩展为 (m , n, 5)
|
||||
dmpr_det = dmpr_det[np.newaxis, ...].repeat(new_yolo_det.shape[0], 0)
|
||||
yolo_dmpr = np.concatenate((new_yolo, dmpr_det), axis=2) # (m, n, 10)
|
||||
|
||||
logger.info("膨胀后标记 :{}, yolo_dmpr:{}", '+++' * 10, yolo_dmpr)
|
||||
x_p, y_p = yolo_dmpr[..., 8], yolo_dmpr[..., 9]
|
||||
x1, y1, x2, y2 = yolo_dmpr[..., 0], yolo_dmpr[..., 1], yolo_dmpr[..., 2], yolo_dmpr[..., 3]
|
||||
x_c, y_c = yolo_dmpr[..., 5], yolo_dmpr[..., 6]
|
||||
|
||||
direction1 = np.arctan2(y_c - y_p, x_c - x_p) / math.pi * 180
|
||||
direction2 = yolo_dmpr[..., 10] / math.pi * 180
|
||||
direction3 = direction2 + 90 # L形角点另外一个方向
|
||||
direction3[direction3 > 180] -= 360
|
||||
ang_diff = direction1 - direction2
|
||||
ang_diff2 = direction1 - direction3
|
||||
|
||||
# 判断膨胀后yolo框包含角点关系 && 包含角点的时候计算水平框中心点与角点的角度关系
|
||||
# direction ∈ （-180， 180） 若角差大于180，需算补角
|
||||
# T形角点比较一个方向，L形角点比较两个方向
|
||||
mask = (x_p >= x1) & (x_p <= x2) & (y_p >= y1) & (y_p <= y2) & \
|
||||
(((yolo_dmpr[..., 11] <= 0.5) & # T形角点情况
|
||||
(((ang_diff >= -90) & (ang_diff <= 90)) | ((ang_diff > 180) & ((360 - ang_diff) <= 90)) |
|
||||
(((ang_diff) < -180) & ((360 + ang_diff) <= 90)))) |
|
||||
((yolo_dmpr[..., 11] > 0.5) & # L形角点情况
|
||||
(((ang_diff >= -90) & (ang_diff <= 90)) | ((ang_diff > 180) & ((360 - ang_diff) <= 90)) |
|
||||
(((ang_diff) < -180) & ((360 + ang_diff) <= 90))) &
|
||||
(((ang_diff2 >= -90) & (ang_diff2 <= 90)) | ((ang_diff2 > 180) & ((360 - ang_diff2) <= 90)) |
|
||||
(((ang_diff2) < -180) & ((360 + ang_diff2) <= 90)))))
|
||||
logger.info("省略 :{}, mask1:{}", '+++' * 10, mask)
|
||||
res = np.sum(mask, axis=1)
|
||||
logger.info("省略 :{}, mask2:{}", '+++' * 10, res)
|
||||
logger.info("省略 :{}, yolo_det_clone:{}", '+++' * 10, yolo_det_clone)
|
||||
yolo_det_clone[yolo_det_clone[:, -2] == cls, -1] = res
|
||||
#print('##line69 original yolo_det_clone:',yolo_det_clone)
|
||||
#yolo_det_clone[yolo_det_clone[:, -1] == 0, -2] = illParkCls
|
||||
logger.info("省略:{}, yolo_det_clone:{}", '+++' * 10, yolo_det_clone)
|
||||
#print('-'*20,'--line78',yolo_det_clone)
|
||||
yolo_det_clone[ np.logical_and( yolo_det_clone[:,-1]==0,yolo_det_clone[:,-2]==cls) ,-2 ] = illParkCls
|
||||
#print('-'*20,'--line80:',yolo_det_clone)
|
||||
yolo_det_clone = yolo_det_clone[:,0:6]
|
||||
time2=time.time()
|
||||
|
||||
return np.array(yolo_det_clone), 'dmpr_yolo:%.1f'%( (time2-time1)*1000 )
|
||||
def stdc_yolo(stdc_det, yolo_det):
|
||||
im = np.uint8(stdc_det)
|
||||
x_c = ((yolo_det[:, 0] + yolo_det[:, 2]) // 2).astype(int)
|
||||
y_c = ((yolo_det[:, 1] + yolo_det[:, 3]) // 2).astype(int)
|
||||
|
||||
yolo_filted = yolo_det[im[y_c, x_c] == 0]
|
||||
|
||||
return yolo_filted
|
||||
|
||||
def dmpr_yolo_stdc(predsList,pars):
|
||||
if len(predsList)==2:
|
||||
yolo_det, dmpr_det = predsList[0:2]
|
||||
else:
|
||||
yolo_det, dmpr_det,stdc_det = predsList[0:3]
|
||||
if len(yolo_det)==0:
|
||||
return yolo_det,' No yolo detections'
|
||||
if isinstance(yolo_det,list):
|
||||
yolo_det = np.array(yolo_det)
|
||||
if len(predsList)>2:
|
||||
yolo_det = stdc_yolo(stdc_det, yolo_det)
|
||||
return dmpr_yolo(yolo_det, dmpr_det,pars)
|
||||
|
|
@ -0,0 +1,2 @@
|
|||
"""Network model related package."""
|
||||
from .detector import DirectionalPointDetector
|
||||
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
|
|
@ -0,0 +1,64 @@
|
|||
"""Defines the detector network structure."""
|
||||
import torch
|
||||
from torch import nn
|
||||
from DMPRUtils.model.network import define_halve_unit, define_detector_block
|
||||
|
||||
|
||||
class YetAnotherDarknet(nn.modules.Module):
|
||||
"""Yet another darknet, imitating darknet-53 with depth of darknet-19."""
|
||||
def __init__(self, input_channel_size, depth_factor):
|
||||
super(YetAnotherDarknet, self).__init__()
|
||||
layers = []
|
||||
# 0
|
||||
layers += [nn.Conv2d(input_channel_size, depth_factor, kernel_size=3,
|
||||
stride=1, padding=1, bias=False)]
|
||||
layers += [nn.BatchNorm2d(depth_factor)]
|
||||
layers += [nn.LeakyReLU(0.1)]
|
||||
# 1
|
||||
layers += define_halve_unit(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
# 2
|
||||
depth_factor *= 2
|
||||
layers += define_halve_unit(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
# 3
|
||||
depth_factor *= 2
|
||||
layers += define_halve_unit(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
# 4
|
||||
depth_factor *= 2
|
||||
layers += define_halve_unit(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
# 5
|
||||
depth_factor *= 2
|
||||
layers += define_halve_unit(depth_factor)
|
||||
layers += define_detector_block(depth_factor)
|
||||
self.model = nn.Sequential(*layers)
|
||||
|
||||
def forward(self, *x):
|
||||
return self.model(x[0])
|
||||
|
||||
|
||||
class DirectionalPointDetector(nn.modules.Module):
|
||||
"""Detector for point with direction."""
|
||||
def __init__(self, input_channel_size, depth_factor, output_channel_size):
|
||||
super(DirectionalPointDetector, self).__init__()
|
||||
self.extract_feature = YetAnotherDarknet(input_channel_size,
|
||||
depth_factor)
|
||||
layers = []
|
||||
layers += define_detector_block(16 * depth_factor)
|
||||
layers += define_detector_block(16 * depth_factor)
|
||||
layers += [nn.Conv2d(32 * depth_factor, output_channel_size,
|
||||
kernel_size=1, stride=1, padding=0, bias=False)]
|
||||
self.predict = nn.Sequential(*layers)
|
||||
|
||||
def forward(self, *x):
|
||||
prediction = self.predict(self.extract_feature(x[0]))
|
||||
# 4 represents that there are 4 value: confidence, shape, offset_x,
|
||||
# offset_y, whose range is between [0, 1].
|
||||
point_pred, angle_pred = torch.split(prediction, 4, dim=1)
|
||||
point_pred = torch.sigmoid(point_pred)
|
||||
angle_pred = torch.tanh(angle_pred)
|
||||
return torch.cat((point_pred, angle_pred), dim=1)
|
||||
|
|
@ -0,0 +1,54 @@
|
|||
"""Universal network struture unit definition."""
|
||||
from torch import nn
|
||||
|
||||
|
||||
def define_squeeze_unit(basic_channel_size):
|
||||
"""Define a 1x1 squeeze convolution with norm and activation."""
|
||||
conv = nn.Conv2d(2 * basic_channel_size, basic_channel_size, kernel_size=1,
|
||||
stride=1, padding=0, bias=False)
|
||||
norm = nn.BatchNorm2d(basic_channel_size)
|
||||
relu = nn.LeakyReLU(0.1)
|
||||
layers = [conv, norm, relu]
|
||||
return layers
|
||||
|
||||
|
||||
def define_expand_unit(basic_channel_size):
|
||||
"""Define a 3x3 expand convolution with norm and activation."""
|
||||
conv = nn.Conv2d(basic_channel_size, 2 * basic_channel_size, kernel_size=3,
|
||||
stride=1, padding=1, bias=False)
|
||||
norm = nn.BatchNorm2d(2 * basic_channel_size)
|
||||
relu = nn.LeakyReLU(0.1)
|
||||
layers = [conv, norm, relu]
|
||||
return layers
|
||||
|
||||
|
||||
def define_halve_unit(basic_channel_size):
|
||||
"""Define a 4x4 stride 2 expand convolution with norm and activation."""
|
||||
conv = nn.Conv2d(basic_channel_size, 2 * basic_channel_size, kernel_size=4,
|
||||
stride=2, padding=1, bias=False)
|
||||
norm = nn.BatchNorm2d(2 * basic_channel_size)
|
||||
relu = nn.LeakyReLU(0.1)
|
||||
layers = [conv, norm, relu]
|
||||
return layers
|
||||
|
||||
|
||||
def define_depthwise_expand_unit(basic_channel_size):
|
||||
"""Define a 3x3 expand convolution with norm and activation."""
|
||||
conv1 = nn.Conv2d(basic_channel_size, 2 * basic_channel_size,
|
||||
kernel_size=1, stride=1, padding=0, bias=False)
|
||||
norm1 = nn.BatchNorm2d(2 * basic_channel_size)
|
||||
relu1 = nn.LeakyReLU(0.1)
|
||||
conv2 = nn.Conv2d(2 * basic_channel_size, 2 * basic_channel_size, kernel_size=3,
|
||||
stride=1, padding=1, bias=False, groups=2 * basic_channel_size)
|
||||
norm2 = nn.BatchNorm2d(2 * basic_channel_size)
|
||||
relu2 = nn.LeakyReLU(0.1)
|
||||
layers = [conv1, norm1, relu1, conv2, norm2, relu2]
|
||||
return layers
|
||||
|
||||
|
||||
def define_detector_block(basic_channel_size):
|
||||
"""Define a unit composite of a squeeze and expand unit."""
|
||||
layers = []
|
||||
layers += define_squeeze_unit(basic_channel_size)
|
||||
layers += define_expand_unit(basic_channel_size)
|
||||
return layers
|
||||
|
|
@ -0,0 +1,47 @@
|
|||
import os
|
||||
import time,argparse
|
||||
import cv2
|
||||
import torch
|
||||
import sys
|
||||
sys.path.extend(['..' ])
|
||||
from DMPRUtils.model.detector import DirectionalPointDetector
|
||||
from pathlib import Path
|
||||
from segutils.trtUtils import toONNX,ONNXtoTrt
|
||||
from DMPRUtils.yolo_net import Model
|
||||
|
||||
def main(opt):
|
||||
|
||||
pars={'depth_factor':32,'NUM_FEATURE_MAP_CHANNEL':6,'dmpr_thresh':0.3, 'dmprimg_size':640,
|
||||
'mWidth':640,'mHeight':640
|
||||
}
|
||||
|
||||
##以下参数目前不可改
|
||||
#DMPRweights = "weights/urbanManagement/DMPR/dp_detector_499.pth"
|
||||
|
||||
DMPRweights = opt.weights.strip()
|
||||
DMPR_pthFile = Path(DMPRweights)
|
||||
inputShape =(1, 3, pars['mHeight'],pars['mWidth'])#(bs,channels,height,width)
|
||||
DMPR_onnxFile = str(DMPR_pthFile.with_suffix('.onnx'))
|
||||
DMPR_trtFile = DMPR_onnxFile.replace('.onnx','.engine' )
|
||||
|
||||
|
||||
##加载模型，准备好显示字符
|
||||
device = 'cuda:0'
|
||||
|
||||
# DMPR model
|
||||
#DMPRmodel = DirectionalPointDetector(3, pars['depth_factor'], pars['NUM_FEATURE_MAP_CHANNEL']).to(device)
|
||||
confUrl = os.path.join( os.path.dirname(__file__),'config','yolov5s.yaml' )
|
||||
DMPRmodel = Model(confUrl, ch=3).to(device)
|
||||
|
||||
|
||||
DMPRmodel.load_state_dict(torch.load(DMPRweights))
|
||||
|
||||
toONNX(DMPRmodel,DMPR_onnxFile,inputShape=inputShape,device=device,dynamic=True)
|
||||
ONNXtoTrt(DMPR_onnxFile,DMPR_trtFile)
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--weights', type=str, default='/mnt/thsw2/DSP2/weights/cityMangement2/weights/urbanManagement/DMPR/dp_detector_499.pth', help='model path(s)')
|
||||
opt = parser.parse_args()
|
||||
|
||||
main(opt)
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
weights=/mnt/thsw2/DSP2/weights/cityMangement3/dmpr
|
||||
#weights=/mnt/thsw2/DSP2/weights/cityMangement2_0916/weights/urbanManagement/DMPR/dp_detector_299
|
||||
gpu=2080Ti
|
||||
python toTrt.py --weights ${weights}.pth
|
||||
mv ${weights}.engine ${weights}_${gpu}.engine
|
||||
|
|
@ -0,0 +1,284 @@
|
|||
# YOLOv5 YOLO-specific modules
|
||||
|
||||
import argparse
|
||||
import logging
|
||||
import sys
|
||||
from copy import deepcopy
|
||||
|
||||
import torch
|
||||
|
||||
sys.path.append('./') # to run '$ python *.py' files in subdirectories
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
from models.common import *
|
||||
from models.experimental import *
|
||||
from utils.autoanchor import check_anchor_order
|
||||
from utils.general import make_divisible, check_file, set_logging
|
||||
from utils.torch_utils import time_synchronized, fuse_conv_and_bn, model_info, scale_img, initialize_weights, \
|
||||
select_device, copy_attr
|
||||
|
||||
try:
|
||||
import thop # for FLOPS computation
|
||||
except ImportError:
|
||||
thop = None
|
||||
|
||||
|
||||
class Detect(nn.Module):
|
||||
stride = None # strides computed during build
|
||||
export = False # onnx export
|
||||
|
||||
def __init__(self, nc=80, anchors=(), ch=()): # detection layers
|
||||
super(Detect, self).__init__()
|
||||
self.no = 6
|
||||
self.nl = 3
|
||||
self.na = len(anchors[0]) // 2 # number of anchors
|
||||
self.grid = [torch.zeros(1)] * self.nl # init grid
|
||||
a = torch.tensor(anchors).float().view(self.nl, -1, 2)
|
||||
self.register_buffer('anchors', a) # shape(nl,na,2)
|
||||
self.register_buffer('anchor_grid', a.clone().view(self.nl, 1, -1, 1, 1, 2)) # shape(nl,1,na,1,1,2)
|
||||
self.m = nn.ModuleList(nn.Conv2d(x, self.no, 1) for x in ch) # output conv
|
||||
|
||||
def forward(self, x):
|
||||
# x = x.copy() # for profiling
|
||||
# z = [] # inference output
|
||||
# # self.training |= self.export
|
||||
# for i in range(self.nl):
|
||||
# x[i] = self.m[i](x[i]) # conv
|
||||
# bs, _, ny, nx = x[i].shape # x(bs,255,20,20) to x(bs,3,20,20,85)
|
||||
# x[i] = x[i].view(bs, self.na, self.no, ny, nx).permute(0, 1, 3, 4, 2).contiguous()
|
||||
#
|
||||
# if not self.training: # inference
|
||||
# if self.grid[i].shape[2:4] != x[i].shape[2:4]:
|
||||
# self.grid[i] = self._make_grid(nx, ny).to(x[i].device)
|
||||
#
|
||||
# y = x[i].sigmoid()
|
||||
# y[..., 0:2] = (y[..., 0:2] * 2. - 0.5 + self.grid[i]) * self.stride[i] # xy
|
||||
# y[..., 2:4] = (y[..., 2:4] * 2) ** 2 * self.anchor_grid[i] # wh
|
||||
# z.append(y.view(bs, -1, self.no))
|
||||
|
||||
prediction = self.m[1](x[1])
|
||||
point_pred, angle_pred = torch.split(prediction, 4, dim=1)
|
||||
point_pred = torch.sigmoid(point_pred)
|
||||
angle_pred = torch.tanh(angle_pred)
|
||||
|
||||
return torch.cat((point_pred, angle_pred), dim=1)
|
||||
|
||||
@staticmethod
|
||||
def _make_grid(nx=20, ny=20):
|
||||
yv, xv = torch.meshgrid([torch.arange(ny), torch.arange(nx)])
|
||||
return torch.stack((xv, yv), 2).view((1, 1, ny, nx, 2)).float()
|
||||
|
||||
|
||||
class Model(nn.Module):
|
||||
def __init__(self, cfg='yolov5s.yaml', ch=3, nc=None, anchors=None): # model, input channels, number of classes
|
||||
super(Model, self).__init__()
|
||||
if isinstance(cfg, dict):
|
||||
self.yaml = cfg # model dict
|
||||
else: # is *.yaml
|
||||
import yaml # for torch hub
|
||||
self.yaml_file = Path(cfg).name
|
||||
with open(cfg) as f:
|
||||
self.yaml = yaml.load(f, Loader=yaml.SafeLoader) # model dict
|
||||
|
||||
# Define model
|
||||
ch = self.yaml['ch'] = self.yaml.get('ch', ch) # input channels
|
||||
if nc and nc != self.yaml['nc']:
|
||||
logger.info(f"Overriding model.yaml nc={self.yaml['nc']} with nc={nc}")
|
||||
self.yaml['nc'] = nc # override yaml value
|
||||
if anchors:
|
||||
logger.info(f'Overriding model.yaml anchors with anchors={anchors}')
|
||||
self.yaml['anchors'] = round(anchors) # override yaml value
|
||||
self.model, self.save = parse_model(deepcopy(self.yaml), ch=[ch]) # model, savelist
|
||||
self.names = [str(i) for i in range(self.yaml['nc'])] # default names
|
||||
# print([x.shape for x in self.forward(torch.zeros(1, ch, 64, 64))])
|
||||
|
||||
# Build strides, anchors
|
||||
# m = self.model[-1] # Detect()
|
||||
# if isinstance(m, Detect):
|
||||
# s = 256 # 2x min stride
|
||||
# m.stride = torch.tensor([s / x.shape[-2] for x in self.forward(torch.zeros(1, ch, s, s))]) # forward
|
||||
# m.anchors /= m.stride.view(-1, 1, 1)
|
||||
# check_anchor_order(m)
|
||||
# self.stride = m.stride
|
||||
# self._initialize_biases() # only run once
|
||||
# print('Strides: %s' % m.stride.tolist())
|
||||
|
||||
# Init weights, biases
|
||||
initialize_weights(self)
|
||||
self.info()
|
||||
logger.info('')
|
||||
|
||||
def forward(self, x, augment=False, profile=False):
|
||||
if augment:
|
||||
img_size = x.shape[-2:] # height, width
|
||||
s = [1, 0.83, 0.67] # scales
|
||||
f = [None, 3, None] # flips (2-ud, 3-lr)
|
||||
y = [] # outputs
|
||||
for si, fi in zip(s, f):
|
||||
xi = scale_img(x.flip(fi) if fi else x, si, gs=int(self.stride.max()))
|
||||
yi = self.forward_once(xi)[0] # forward
|
||||
# cv2.imwrite(f'img_{si}.jpg', 255 * xi[0].cpu().numpy().transpose((1, 2, 0))[:, :, ::-1]) # save
|
||||
yi[..., :4] /= si # de-scale
|
||||
if fi == 2:
|
||||
yi[..., 1] = img_size[0] - yi[..., 1] # de-flip ud
|
||||
elif fi == 3:
|
||||
yi[..., 0] = img_size[1] - yi[..., 0] # de-flip lr
|
||||
y.append(yi)
|
||||
return torch.cat(y, 1), None # augmented inference, train
|
||||
else:
|
||||
return self.forward_once(x, profile) # single-scale inference, train
|
||||
|
||||
def forward_once(self, x, profile=False):
|
||||
y, dt = [], [] # outputs
|
||||
for m in self.model:
|
||||
if m.f != -1: # if not from previous layer
|
||||
x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f] # from earlier layers
|
||||
|
||||
if profile:
|
||||
o = thop.profile(m, inputs=(x,), verbose=False)[0] / 1E9 * 2 if thop else 0 # FLOPS
|
||||
t = time_synchronized()
|
||||
for _ in range(10):
|
||||
_ = m(x)
|
||||
dt.append((time_synchronized() - t) * 100)
|
||||
print('%10.1f%10.0f%10.1fms %-40s' % (o, m.np, dt[-1], m.type))
|
||||
|
||||
x = m(x) # run
|
||||
y.append(x if m.i in self.save else None) # save output
|
||||
|
||||
if profile:
|
||||
print('%.1fms total' % sum(dt))
|
||||
return x
|
||||
|
||||
def _initialize_biases(self, cf=None): # initialize biases into Detect(), cf is class frequency
|
||||
# https://arxiv.org/abs/1708.02002 section 3.3
|
||||
# cf = torch.bincount(torch.tensor(np.concatenate(dataset.labels, 0)[:, 0]).long(), minlength=nc) + 1.
|
||||
m = self.model[-1] # Detect() module
|
||||
for mi, s in zip(m.m, m.stride): # from
|
||||
b = mi.bias.view(m.na, -1) # conv.bias(255) to (3,85)
|
||||
b.data[:, 4] += math.log(8 / (640 / s) ** 2) # obj (8 objects per 640 image)
|
||||
b.data[:, 5:] += math.log(0.6 / (m.nc - 0.99)) if cf is None else torch.log(cf / cf.sum()) # cls
|
||||
mi.bias = torch.nn.Parameter(b.view(-1), requires_grad=True)
|
||||
|
||||
def _print_biases(self):
|
||||
m = self.model[-1] # Detect() module
|
||||
for mi in m.m: # from
|
||||
b = mi.bias.detach().view(m.na, -1).T # conv.bias(255) to (3,85)
|
||||
print(('%6g Conv2d.bias:' + '%10.3g' * 6) % (mi.weight.shape[1], *b[:5].mean(1).tolist(), b[5:].mean()))
|
||||
|
||||
# def _print_weights(self):
|
||||
# for m in self.model.modules():
|
||||
# if type(m) is Bottleneck:
|
||||
# print('%10.3g' % (m.w.detach().sigmoid() * 2)) # shortcut weights
|
||||
|
||||
def fuse(self): # fuse model Conv2d() + BatchNorm2d() layers
|
||||
print('Fusing layers... ')
|
||||
for m in self.model.modules():
|
||||
if type(m) is Conv and hasattr(m, 'bn'):
|
||||
m.conv = fuse_conv_and_bn(m.conv, m.bn) # update conv
|
||||
delattr(m, 'bn') # remove batchnorm
|
||||
m.forward = m.fuseforward # update forward
|
||||
self.info()
|
||||
return self
|
||||
|
||||
def nms(self, mode=True): # add or remove NMS module
|
||||
present = type(self.model[-1]) is NMS # last layer is NMS
|
||||
if mode and not present:
|
||||
print('Adding NMS... ')
|
||||
m = NMS() # module
|
||||
m.f = -1 # from
|
||||
m.i = self.model[-1].i + 1 # index
|
||||
self.model.add_module(name='%s' % m.i, module=m) # add
|
||||
self.eval()
|
||||
elif not mode and present:
|
||||
print('Removing NMS... ')
|
||||
self.model = self.model[:-1] # remove
|
||||
return self
|
||||
|
||||
def autoshape(self): # add autoShape module
|
||||
print('Adding autoShape... ')
|
||||
m = autoShape(self) # wrap model
|
||||
copy_attr(m, self, include=('yaml', 'nc', 'hyp', 'names', 'stride'), exclude=()) # copy attributes
|
||||
return m
|
||||
|
||||
def info(self, verbose=False, img_size=640): # print model information
|
||||
model_info(self, verbose, img_size)
|
||||
|
||||
|
||||
def parse_model(d, ch): # model_dict, input_channels(3)
|
||||
logger.info('\n%3s%18s%3s%10s %-40s%-30s' % ('', 'from', 'n', 'params', 'module', 'arguments'))
|
||||
anchors, nc, gd, gw = d['anchors'], d['nc'], d['depth_multiple'], d['width_multiple']
|
||||
na = (len(anchors[0]) // 2) if isinstance(anchors, list) else anchors # number of anchors
|
||||
# no = na * (nc + 5) # number of outputs = anchors * (classes + 5)
|
||||
no = 6
|
||||
|
||||
layers, save, c2 = [], [], ch[-1] # layers, savelist, ch out
|
||||
for i, (f, n, m, args) in enumerate(d['backbone'] + d['head']): # from, number, module, args
|
||||
m = eval(m) if isinstance(m, str) else m # eval strings
|
||||
for j, a in enumerate(args):
|
||||
try:
|
||||
args[j] = eval(a) if isinstance(a, str) else a # eval strings
|
||||
except:
|
||||
pass
|
||||
|
||||
n = max(round(n * gd), 1) if n > 1 else n # depth gain
|
||||
if m in [Conv, GhostConv, Bottleneck, GhostBottleneck, SPP, DWConv, MixConv2d, Focus, CrossConv, BottleneckCSP,
|
||||
C3, C3TR]:
|
||||
c1, c2 = ch[f], args[0]
|
||||
if c2 != no: # if not output
|
||||
c2 = make_divisible(c2 * gw, 8)
|
||||
|
||||
args = [c1, c2, *args[1:]]
|
||||
if m in [BottleneckCSP, C3, C3TR]:
|
||||
args.insert(2, n) # number of repeats
|
||||
n = 1
|
||||
elif m is nn.BatchNorm2d:
|
||||
args = [ch[f]]
|
||||
elif m is Concat:
|
||||
c2 = sum([ch[x] for x in f])
|
||||
elif m is Detect:
|
||||
args.append([ch[x] for x in f])
|
||||
if isinstance(args[1], int): # number of anchors
|
||||
args[1] = [list(range(args[1] * 2))] * len(f)
|
||||
elif m is Contract:
|
||||
c2 = ch[f] * args[0] ** 2
|
||||
elif m is Expand:
|
||||
c2 = ch[f] // args[0] ** 2
|
||||
else:
|
||||
c2 = ch[f]
|
||||
|
||||
m_ = nn.Sequential(*[m(*args) for _ in range(n)]) if n > 1 else m(*args) # module
|
||||
t = str(m)[8:-2].replace('__main__.', '') # module type
|
||||
np = sum([x.numel() for x in m_.parameters()]) # number params
|
||||
m_.i, m_.f, m_.type, m_.np = i, f, t, np # attach index, 'from' index, type, number params
|
||||
logger.info('%3s%18s%3s%10.0f %-40s%-30s' % (i, f, n, np, t, args)) # print
|
||||
save.extend(x % i for x in ([f] if isinstance(f, int) else f) if x != -1) # append to savelist
|
||||
layers.append(m_)
|
||||
if i == 0:
|
||||
ch = []
|
||||
ch.append(c2)
|
||||
return nn.Sequential(*layers), sorted(save)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
parser = argparse.ArgumentParser()
|
||||
parser.add_argument('--cfg', type=str, default='yolov5s.yaml', help='model.yaml')
|
||||
parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
|
||||
opt = parser.parse_args()
|
||||
opt.cfg = check_file(opt.cfg) # check file
|
||||
set_logging()
|
||||
device = select_device(opt.device)
|
||||
|
||||
# Create model
|
||||
model = Model(opt.cfg).to(device)
|
||||
model.train()
|
||||
|
||||
# Profile
|
||||
# img = torch.rand(8 if torch.cuda.is_available() else 1, 3, 640, 640).to(device)
|
||||
# y = model(img, profile=True)
|
||||
|
||||
# Tensorboard
|
||||
# from torch.utils.tensorboard import SummaryWriter
|
||||
# tb_writer = SummaryWriter()
|
||||
# print("Run 'tensorboard --logdir=models/runs' to view tensorboard at http://localhost:6006/")
|
||||
# tb_writer.add_graph(model.model, img) # add model to tensorboard
|
||||
# tb_writer.add_image('test', img[0], dataformats='CWH') # add model to tensorboard
|
||||
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
Binary file not shown.
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["钓鱼","游泳"],
|
||||
"labelIndexs":["SL01","SL02"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,61 @@
|
|||
GPUID: 0
|
||||
WORKERS: 1
|
||||
PRINT_FREQ: 10
|
||||
SAVE_FREQ: 10
|
||||
PIN_MEMORY: False
|
||||
OUTPUT_DIR: 'output'
|
||||
|
||||
CUDNN:
|
||||
BENCHMARK: True
|
||||
DETERMINISTIC: False
|
||||
ENABLED: True
|
||||
|
||||
DATASET:
|
||||
DATASET: 360CC
|
||||
ROOT: "../textGenerator/dataset/dataset9/images"
|
||||
CHAR_FILE: '../textGenerator/dataset/dataset9/chars.txt'
|
||||
JSON_FILE: {'train': '../textGenerator/dataset/dataset9/train.txt', 'val': '../textGenerator/dataset/dataset9/val.txt'}
|
||||
# JSON_FILE: {'train': 'H:/DL-DATASET/360M/train.txt', 'val': 'H:/DL-DATASET/360M/test.txt'}
|
||||
SCALE_FACTOR: 0.25
|
||||
ROT_FACTOR: 30
|
||||
STD: 0.193
|
||||
MEAN: 0.588
|
||||
ALPHABETS: ''
|
||||
|
||||
TRAIN:
|
||||
BATCH_SIZE_PER_GPU: 32
|
||||
SHUFFLE: True
|
||||
BEGIN_EPOCH: 0
|
||||
END_EPOCH: 100
|
||||
RESUME:
|
||||
IS_RESUME: False
|
||||
FILE: 'output/360CC/crnn/2023-04-27-13-01/checkpoints/checkpoint_99_acc_0.5030.pth'
|
||||
OPTIMIZER: 'adam'
|
||||
LR: 0.0001
|
||||
WD: 0.0
|
||||
LR_STEP: [60, 80]
|
||||
LR_FACTOR: 0.1
|
||||
MOMENTUM: 0.0
|
||||
NESTEROV: False
|
||||
RMSPROP_ALPHA:
|
||||
RMSPROP_CENTERED:
|
||||
FINETUNE:
|
||||
IS_FINETUNE: False
|
||||
FINETUNE_CHECKPOINIT: 'output/checkpoints/mixed_second_finetune_acc_97P7.pth'
|
||||
FREEZE: true
|
||||
|
||||
TEST:
|
||||
BATCH_SIZE_PER_GPU: 16
|
||||
SHUFFLE: True # for random test rather than test on the whole validation set
|
||||
NUM_TEST_BATCH: 1000
|
||||
NUM_TEST_DISP: 10
|
||||
|
||||
MODEL:
|
||||
NAME: 'crnn'
|
||||
IMAGE_SIZE:
|
||||
OW: 160 # origial width: 280
|
||||
H: 32
|
||||
W: 160 # resized width: 160
|
||||
NUM_CLASSES: 0
|
||||
NUM_HIDDEN: 256
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,92 @@
|
|||
a
|
||||
b
|
||||
c
|
||||
d
|
||||
e
|
||||
f
|
||||
g
|
||||
h
|
||||
i
|
||||
j
|
||||
k
|
||||
l
|
||||
m
|
||||
n
|
||||
o
|
||||
p
|
||||
q
|
||||
r
|
||||
s
|
||||
t
|
||||
u
|
||||
v
|
||||
w
|
||||
x
|
||||
y
|
||||
z
|
||||
A
|
||||
B
|
||||
C
|
||||
D
|
||||
E
|
||||
F
|
||||
G
|
||||
H
|
||||
I
|
||||
J
|
||||
K
|
||||
L
|
||||
M
|
||||
N
|
||||
O
|
||||
P
|
||||
Q
|
||||
R
|
||||
S
|
||||
T
|
||||
U
|
||||
V
|
||||
W
|
||||
X
|
||||
Y
|
||||
Z
|
||||
0
|
||||
1
|
||||
2
|
||||
3
|
||||
4
|
||||
5
|
||||
6
|
||||
7
|
||||
8
|
||||
9
|
||||
°
|
||||
!
|
||||
"
|
||||
#
|
||||
$
|
||||
%
|
||||
&
|
||||
'
|
||||
(
|
||||
)
|
||||
*
|
||||
+
|
||||
,
|
||||
-
|
||||
.
|
||||
/
|
||||
:
|
||||
;
|
||||
?
|
||||
@
|
||||
[
|
||||
\
|
||||
]
|
||||
^
|
||||
_
|
||||
`
|
||||
{
|
||||
|
|
||||
}
|
||||
~
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames_实际":["国旗","浮标","船名","船只","未挂国旗船只" ],
|
||||
"labelnames":[ "国旗","浮标","船名","船只","未挂国旗船只" ],
|
||||
"labelIndexs":["SL040", "SL041","SL042","SL043","SL044"]
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["人"],
|
||||
"labelIndexs":["SL031"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,11 @@
|
|||
#检测混合模型
|
||||
for bus in highWay2 river2 drowning noParking river
|
||||
do
|
||||
diff /mnt/thsw2/DSP2/weights/${bus}/yolov5.pt ${bus}/yolov5.pt
|
||||
diff /mnt/thsw2/DSP2/weights/${bus}/stdc_360X640.pth ${bus}/stdc_360X640.pth
|
||||
done
|
||||
#检查检测模型
|
||||
for bus in forest2 vehicle pedestrian smogfire AnglerSwimmer countryRoad cityMangement
|
||||
do
|
||||
diff /mnt/thsw2/DSP2/weights/${bus}/yolov5.pt ${bus}/yolov5.pt
|
||||
done
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["车辆","垃圾","商贩"],
|
||||
"labelIndexs":["SL01","SL02","SL03"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["车辆","垃圾","商贩","违停"],
|
||||
"labelIndexs":["SL01","SL02","SL03","SL04"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"score_byClass":{"0":0.88,"1":0.3,"2":0.3,"3":0.3 } ,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
Binary file not shown.
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["车辆","垃圾","商贩","违停"],
|
||||
"labelIndexs":["SL01","SL02","SL03","SL04"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"score_byClass":{"0":0.88,"1":0.3,"2":0.3,"3":0.3 } ,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames":["护栏","交通标志","非交通标志","施工","施工"],
|
||||
"labelIndexs":["SL01","SL02","SL03","SL04","SL05"],
|
||||
"labelnamesActual":["护栏","交通标志","非交通标志","锥桶","水马" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["违法种植"],
|
||||
"labelIndexs":["SL01"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":[ "纵向裂缝","横向裂缝","网状裂缝" ],
|
||||
"labelIndexs":["SL01","SL02","SL03"]
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
crack_yolov5_202302.pt对应类别['pedestrian', 'vehicle', 'D00', 'D10', 'Repair', 'D20', 'D40', 'Block crack', 'JiShui']
|
||||
roaddamage20231028.pt对应类别[ 'D00','D10','D20','D40','D44','D50','Repair','D43','D01','D11','D0w0','Block crack' ]
|
||||
[ 'D00'：纵向裂缝,
|
||||
'D10'：横向裂缝,
|
||||
'D20'：网状裂缝 ]
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["人头"],
|
||||
"labelIndexs":["SL01"]
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
|
||||
"labelnames":[ "人头","人","船只" ],
|
||||
"labelIndexs":[ "SL001","SL002","SL003" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{
|
||||
"name":"post_process","conf_thres":0.25,"iou_thres":0.25,"classes":9,
|
||||
"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,0],[255,255,0],[255,0,0],[255,0,127],[255,0,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["林斑","病死树"],
|
||||
"labelIndexs":["SL031","SL032"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["林斑","病死树","行人","火焰","烟雾","云朵"],
|
||||
"labelIndexs":["SL031","SL032","SL033","SL034","SL035","SL036","SL037"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["林斑","病死树","行人","火焰","烟雾","人群"],
|
||||
"labelIndexs":["SL031","SL032","SL033","SL034","SL035","SL036","SL037"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
name,r,g,b,cls
|
||||
0,0,0,0,bg
|
||||
1,128,0,0,road
|
||||
2,0,128,0,vehicle
|
||||
|
|
|
@ -0,0 +1,6 @@
|
|||
{
|
||||
|
||||
"labelnames":["行人","车辆","裂缝","裂缝","修补","裂缝","坑槽","裂缝","积水", "影子","事故"],
|
||||
"labelnames_实际":["行人","车辆","纵向裂缝","横向裂缝","修补","网状裂纹","坑槽","块状裂纹","积水","影子","事故"],
|
||||
"labelIndexs":["SL01","SL02","SL03","SL04","SL05","SL06","SL007","SL008","SL009","SL010","SL011" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{
|
||||
"name":"post_process","conf_thres":0.25,"iou_thres":0.25,"classes":9,
|
||||
"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,0],[255,255,0],[255,0,0],[255,0,127],[255,0,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
|
||||
"labelnames":[ "车","T角点","L角点","违停" ],
|
||||
"labelIndexs":[ "SL001","SL002","SL003","SL004" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{
|
||||
"name":"post_process","conf_thres":0.25,"iou_thres":0.25,"classes":9,
|
||||
"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,0],[255,255,0],[255,0,0],[255,0,127],[255,0,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
Binary file not shown.
Binary file not shown.
|
|
@ -0,0 +1,5 @@
|
|||
name,cls
|
||||
背景,0
|
||||
道路,1
|
||||
车道线,2
|
||||
车辆,3
|
||||
|
|
|
@ -0,0 +1,6 @@
|
|||
{
|
||||
|
||||
"labelnames":[ "车辆","违停" ],
|
||||
"labelIndexs":[ "SL001","SL002" ]
|
||||
}
|
||||
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{
|
||||
"name":"post_process","conf_thres":0.25,"iou_thres":0.25,"classes":9,
|
||||
"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,0],[255,255,0],[255,0,0],[255,0,127],[255,0,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,61 @@
|
|||
GPUID: 0
|
||||
WORKERS: 1
|
||||
PRINT_FREQ: 10
|
||||
SAVE_FREQ: 10
|
||||
PIN_MEMORY: False
|
||||
OUTPUT_DIR: 'output'
|
||||
|
||||
CUDNN:
|
||||
BENCHMARK: True
|
||||
DETERMINISTIC: False
|
||||
ENABLED: True
|
||||
|
||||
DATASET:
|
||||
DATASET: 360CC
|
||||
ROOT: "../textGenerator/dataset/dataset9/images"
|
||||
CHAR_FILE: '../textGenerator/dataset/dataset9/chars.txt'
|
||||
JSON_FILE: {'train': '../textGenerator/dataset/dataset9/train.txt', 'val': '../textGenerator/dataset/dataset9/val.txt'}
|
||||
# JSON_FILE: {'train': 'H:/DL-DATASET/360M/train.txt', 'val': 'H:/DL-DATASET/360M/test.txt'}
|
||||
SCALE_FACTOR: 0.25
|
||||
ROT_FACTOR: 30
|
||||
STD: 0.193
|
||||
MEAN: 0.588
|
||||
ALPHABETS: ''
|
||||
|
||||
TRAIN:
|
||||
BATCH_SIZE_PER_GPU: 32
|
||||
SHUFFLE: True
|
||||
BEGIN_EPOCH: 0
|
||||
END_EPOCH: 100
|
||||
RESUME:
|
||||
IS_RESUME: False
|
||||
FILE: 'output/360CC/crnn/2023-04-27-13-01/checkpoints/checkpoint_99_acc_0.5030.pth'
|
||||
OPTIMIZER: 'adam'
|
||||
LR: 0.0001
|
||||
WD: 0.0
|
||||
LR_STEP: [60, 80]
|
||||
LR_FACTOR: 0.1
|
||||
MOMENTUM: 0.0
|
||||
NESTEROV: False
|
||||
RMSPROP_ALPHA:
|
||||
RMSPROP_CENTERED:
|
||||
FINETUNE:
|
||||
IS_FINETUNE: False
|
||||
FINETUNE_CHECKPOINIT: 'output/checkpoints/mixed_second_finetune_acc_97P7.pth'
|
||||
FREEZE: true
|
||||
|
||||
TEST:
|
||||
BATCH_SIZE_PER_GPU: 16
|
||||
SHUFFLE: True # for random test rather than test on the whole validation set
|
||||
NUM_TEST_BATCH: 1000
|
||||
NUM_TEST_DISP: 10
|
||||
|
||||
MODEL:
|
||||
NAME: 'crnn'
|
||||
IMAGE_SIZE:
|
||||
OW: 160 # origial width: 280
|
||||
H: 32
|
||||
W: 160 # resized width: 160
|
||||
NUM_CLASSES: 0
|
||||
NUM_HIDDEN: 256
|
||||
|
||||
File diff suppressed because one or more lines are too long
|
|
@ -0,0 +1,92 @@
|
|||
a
|
||||
b
|
||||
c
|
||||
d
|
||||
e
|
||||
f
|
||||
g
|
||||
h
|
||||
i
|
||||
j
|
||||
k
|
||||
l
|
||||
m
|
||||
n
|
||||
o
|
||||
p
|
||||
q
|
||||
r
|
||||
s
|
||||
t
|
||||
u
|
||||
v
|
||||
w
|
||||
x
|
||||
y
|
||||
z
|
||||
A
|
||||
B
|
||||
C
|
||||
D
|
||||
E
|
||||
F
|
||||
G
|
||||
H
|
||||
I
|
||||
J
|
||||
K
|
||||
L
|
||||
M
|
||||
N
|
||||
O
|
||||
P
|
||||
Q
|
||||
R
|
||||
S
|
||||
T
|
||||
U
|
||||
V
|
||||
W
|
||||
X
|
||||
Y
|
||||
Z
|
||||
0
|
||||
1
|
||||
2
|
||||
3
|
||||
4
|
||||
5
|
||||
6
|
||||
7
|
||||
8
|
||||
9
|
||||
°
|
||||
!
|
||||
"
|
||||
#
|
||||
$
|
||||
%
|
||||
&
|
||||
'
|
||||
(
|
||||
)
|
||||
*
|
||||
+
|
||||
,
|
||||
-
|
||||
.
|
||||
/
|
||||
:
|
||||
;
|
||||
?
|
||||
@
|
||||
[
|
||||
\
|
||||
]
|
||||
^
|
||||
_
|
||||
`
|
||||
{
|
||||
|
|
||||
}
|
||||
~
|
||||
|
|
@ -0,0 +1 @@
|
|||
abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789°!"#$%&'()*+,-./:;?@[\]^_`{|}~
|
||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1 @@
|
|||
0123456789!"#$%&'()*+,-./:;<=>?@[\]^_`{|}~ €ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz
|
||||
|
Can't render this file because it contains an unexpected character in line 1 and column 12.
|
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["行人"],
|
||||
"labelIndexs":["SL01"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
Binary file not shown.
|
|
@ -0,0 +1,6 @@
|
|||
{
|
||||
|
||||
"labelnames":["坑槽"],
|
||||
"labelnames_实际":["坑槽"],
|
||||
"labelIndexs":["SL01" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{
|
||||
"name":"post_process","conf_thres":0.25,"iou_thres":0.25,"classes":9,
|
||||
"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,0],[255,255,0],[255,0,0],[255,0,127],[255,0,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames":["排口","水生植被","其它","漂浮物","污染排口","菜地","违建","岸坡垃圾"],
|
||||
"labelIndexs":["SL04","SL011","SL013","SL001","SL001","SL002","SL003","SL004" ],
|
||||
"labelOrders":[0,1,2,3,4,5,6,7]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames":[ "漂浮物","垃圾","排口","非法建筑","非法种植","水生植物","游泳人员","钓鱼人员","船只","蓝藻"] ,
|
||||
"labelIndexs":[ "SL04","SL05","SL06","SL07","SL08","SL09","SL10","SL11","SL12","SL13" ],
|
||||
"labelOrders":[0,1,2,3,4,5,6,7,8,9]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.3,"ovlap_thres_crossCategory":0.65,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames_实际":["纵向裂缝","横向裂缝","修补","网状裂纹","坑槽","块状裂纹","积水"],
|
||||
"labelnames":["裂缝","裂缝","修补","裂缝","坑槽","裂缝","积水"],
|
||||
"labelIndexs":["SL030","SL031","SL032","SL033","SL034","SL035","SL036"]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,
|
||||
"rainbows":[[0,0,255],[0,0,255],[255,0,0],[0,0,255],[255,255,0],[0,0,255],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"labelnames":["船只"],
|
||||
"labelIndexs":["SL01"]
|
||||
}
|
||||
|
|
@ -0,0 +1,7 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,"rainbows":[ [0,0,255],[0,255,0],[255,0,0],[255,0,255],[255,255,0],[255,129,0],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]] }
|
||||
|
||||
|
||||
}
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
name,r,g,b,cls
|
||||
0,0,0,0,bg
|
||||
1,128,0,0,road
|
||||
2,0,128,0,vehicle
|
||||
|
|
|
@ -0,0 +1,5 @@
|
|||
{
|
||||
"labelnames_实际":[ "0","1","2","3","4","5","6","7","8","9","10","11","12","13","boat" ],
|
||||
"labelnames":[ "0","1","2","3","4","5","6","7","8","9","10","11","12","13","船只" ],
|
||||
"labelIndexs":["SL050", "SL051", "SL052", "SL053", "SL054", "SL055", "SL056", "SL057", "SL058", "SL059", "SL060", "SL061", "SL062","SL063", "SL064" ]
|
||||
}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{
|
||||
|
||||
|
||||
"post_process":{ "name":"post_process","conf_thres":0.25,"iou_thres":0.45,"classes":5,
|
||||
"rainbows":[[0,0,255],[0,0,255],[255,0,0],[0,0,255],[255,255,0],[0,0,255],[255,0,127],[127,255,0],[0,255,127],[0,127,255],[127,0,255],[255,127,255],[255,255,127],[127,255,255],[0,255,255],[255,127,255],[127,255,255], [0,127,0],[0,0,127],[0,255,255]]
|
||||
|
||||
}
|
||||
|
||||
|
||||
}
|
||||
Some files were not shown because too many files have changed in this diff Show More
Loading…
Reference in New Issue