|
|
@@ -1,39 +1,35 @@ |
|
|
|
import math |
|
|
|
|
|
|
|
import numpy as np |
|
|
|
import torch |
|
|
|
|
|
|
|
|
|
|
|
def dmpr_yolo(dmpr_det, yolo_det, img_shape, cls:int): |
|
|
|
device_ = yolo_det.device |
|
|
|
|
|
|
|
# dmpr_det内容为conf, x, y, θ, shape |
|
|
|
if dmpr_det.device != device_: |
|
|
|
dmpr_det = dmpr_det.to(device_) |
|
|
|
|
|
|
|
# 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.clone().detach() |
|
|
|
tmp_0_tensor = torch.zeros([len(yolo_det), 1], device=device_) |
|
|
|
yolo_det_clone = torch.cat([yolo_det_clone, tmp_0_tensor], dim=1) |
|
|
|
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 = torch.zeros([len(yolo_det), 5], device=device_) |
|
|
|
new_yolo_det = np.zeros([len(yolo_det), 5]) |
|
|
|
|
|
|
|
# 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).int() + 1)*0.2 |
|
|
|
y_dilate_coefficient = ((~(x_length > y_length)).int() + 1)*0.2 |
|
|
|
x_dilate_coefficient = ((x_length > y_length) + 1)*0.2 |
|
|
|
y_dilate_coefficient = ((~(x_length > y_length)) + 1)*0.2 |
|
|
|
|
|
|
|
# 膨胀 |
|
|
|
new_yolo_det[:, 0] = torch.round(yolo_det[:, 0] - x_dilate_coefficient * x_length).clamp_(0, img_shape[1]) #x1 膨胀 |
|
|
|
new_yolo_det[:, 1] = torch.round(yolo_det[:, 1] - y_dilate_coefficient * y_length).clamp_(0, img_shape[0]) #y1 膨胀 |
|
|
|
new_yolo_det[:, 2] = torch.round(yolo_det[:, 2] + x_dilate_coefficient * x_length).clamp_(0, img_shape[1]) #x2 膨胀 |
|
|
|
new_yolo_det[:, 3] = torch.round(yolo_det[:, 3] + y_dilate_coefficient * y_length).clamp_(0, img_shape[0]) #y2 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 |
|
|
|
|
|
|
|
# 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 |
|
|
|
# for i in range(0, len(new_yolo_det)): |
|
|
@@ -54,19 +50,19 @@ def dmpr_yolo(dmpr_det, yolo_det, img_shape, cls:int): |
|
|
|
# new_yolo_det[i, 4] = j + 1 |
|
|
|
# elif (ang_diff < -180) and (360 + ang_diff <= 90): |
|
|
|
# new_yolo_det[i, 4] = j + 1 |
|
|
|
m, n = len(new_yolo_det), len(dmpr_det) |
|
|
|
m, n = new_yolo_det.size, dmpr_det.size |
|
|
|
if not m or not n: |
|
|
|
return yolo_det_clone, new_yolo_det |
|
|
|
|
|
|
|
new_yolo = new_yolo_det.unsqueeze(dim=1).repeat(1, n, 1) # 扩展为 (m , n, 5) |
|
|
|
dmpr_det = dmpr_det.unsqueeze(dim=0).repeat(m, 1, 1) |
|
|
|
yolo_dmpr = torch.cat((new_yolo, dmpr_det), dim=2) # (m, n, 10) |
|
|
|
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[..., 6], yolo_dmpr[..., 7] |
|
|
|
x1, y1, x2, y2 = yolo_dmpr[..., 0], yolo_dmpr[..., 1], yolo_dmpr[..., 2], yolo_dmpr[..., 3] |
|
|
|
x_c, y_c = (x1+x2)/2, (y1+y2)/2 |
|
|
|
|
|
|
|
direction1 = torch.atan2(y_c - y_p, x_c - x_p) / math.pi * 180 |
|
|
|
direction1 = np.arctan2(y_c - y_p, x_c - x_p) / math.pi * 180 |
|
|
|
direction2 = yolo_dmpr[..., 8] / math.pi * 180 |
|
|
|
ang_diff = direction1 - direction2 |
|
|
|
|
|
|
@@ -75,7 +71,7 @@ def dmpr_yolo(dmpr_det, yolo_det, img_shape, cls:int): |
|
|
|
mask = (x_p >= x1) & (x_p <= x2) & (y_p >= y1) & (y_p <= y2) & \ |
|
|
|
(((ang_diff >= -90) & (ang_diff <= 90)) | ((ang_diff > 180) & ((360 - ang_diff) <= 90)) | (((ang_diff) < -180) & ((360 + ang_diff) <= 90))) |
|
|
|
|
|
|
|
res = torch.sum(mask, dim=1).float() |
|
|
|
res = np.sum(mask, axis=1) |
|
|
|
|
|
|
|
# 索引两次更新tensor test1 |
|
|
|
# yolo_det_clone[yolo_det_clone[:, -2] == cls][:, -1] = new_yolo_det[:, 4] |