Nelze vybrat více než 25 témat Téma musí začínat písmenem nebo číslem, může obsahovat pomlčky („-“) a může být dlouhé až 35 znaků.

před 4 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
před 4 roky
před 4 roky
před 4 roky
před 4 roky
Add TensorFlow and TFLite export (#1127) * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * Put representative dataset in tfl_int8 block * detect.py TF inference * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * Add models/tf.py for TensorFlow and TFLite export * Set auto=False for int8 calibration * Update requirements.txt for TensorFlow and TFLite export * Read anchors directly from PyTorch weights * Add --tf-nms to append NMS in TensorFlow SavedModel and GraphDef export * Remove check_anchor_order, check_file, set_logging from import * Reformat code and optimize imports * Autodownload model and check cfg * update --source path, img-size to 320, single output * Adjust representative_dataset * detect.py TF inference * Put representative dataset in tfl_int8 block * weights to string * weights to string * cleanup tf.py * Add --dynamic-batch-size * Add xywh normalization to reduce calibration error * Update requirements.txt TensorFlow 2.3.1 -> 2.4.0 to avoid int8 quantization error * Fix imports Move C3 from models.experimental to models.common * implement C3() and SiLU() * Fix reshape dim to support dynamic batching * Add epsilon argument in tf_BN, which is different between TF and PT * Set stride to None if not using PyTorch, and do not warmup without PyTorch * Add list support in check_img_size() * Add list input support in detect.py * sys.path.append('./') to run from yolov5/ * Add int8 quantization support for TensorFlow 2.5 * Add get_coco128.sh * Remove --no-tfl-detect in models/tf.py (Use tf-android-tfl-detect branch for EdgeTPU) * Update requirements.txt * Replace torch.load() with attempt_load() * Update requirements.txt * Add --tf-raw-resize to set half_pixel_centers=False * Add --agnostic-nms for TF class-agnostic NMS * Cleanup after merge * Cleanup2 after merge * Cleanup3 after merge * Add tf.py docstring with credit and usage * pb saved_model and tflite use only one model in detect.py * Add use cases in docstring of tf.py * Remove redundant `stride` definition * Remove keras direct import * Fix `check_requirements(('tensorflow>=2.4.1',))` Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>
před 3 roky
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317
  1. # YOLOv5 🚀 by Ultralytics, GPL-3.0 license
  2. """
  3. Run inference on images, videos, directories, streams, etc.
  4. Usage:
  5. $ python path/to/detect.py --source path/to/img.jpg --weights yolov5s.pt --img 640
  6. """
  7. import argparse
  8. import os
  9. import platform
  10. import sys
  11. from pathlib import Path
  12. import cv2
  13. import numpy as np
  14. import torch
  15. import torch.backends.cudnn as cudnn
  16. FILE = Path(__file__).resolve()
  17. ROOT = FILE.parents[0] # YOLOv5 root directory
  18. if str(ROOT) not in sys.path:
  19. sys.path.append(str(ROOT)) # add ROOT to PATH
  20. ROOT = Path(os.path.relpath(ROOT, Path.cwd())) # relative
  21. from models.experimental import attempt_load
  22. from utils.datasets import IMG_FORMATS, VID_FORMATS, LoadImages, LoadStreams
  23. from utils.general import (LOGGER, apply_classifier, check_file, check_img_size, check_imshow, check_requirements,
  24. check_suffix, colorstr, increment_path, non_max_suppression, print_args, save_one_box,
  25. scale_coords, strip_optimizer, xyxy2xywh)
  26. from utils.plots import Annotator, colors
  27. from utils.torch_utils import load_classifier, select_device, time_sync
  28. @torch.no_grad()
  29. def run(weights=ROOT / 'yolov5s.pt', # model.pt path(s)
  30. source=ROOT / 'data/images', # file/dir/URL/glob, 0 for webcam
  31. imgsz=640, # inference size (pixels)
  32. conf_thres=0.25, # confidence threshold
  33. iou_thres=0.45, # NMS IOU threshold
  34. max_det=1000, # maximum detections per image
  35. device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
  36. view_img=False, # show results
  37. save_txt=False, # save results to *.txt
  38. save_conf=False, # save confidences in --save-txt labels
  39. save_crop=False, # save cropped prediction boxes
  40. nosave=False, # do not save images/videos
  41. classes=None, # filter by class: --class 0, or --class 0 2 3
  42. agnostic_nms=False, # class-agnostic NMS
  43. augment=False, # augmented inference
  44. visualize=False, # visualize features
  45. update=False, # update all models
  46. project=ROOT / 'runs/detect', # save results to project/name
  47. name='exp', # save results to project/name
  48. exist_ok=False, # existing project/name ok, do not increment
  49. line_thickness=3, # bounding box thickness (pixels)
  50. hide_labels=False, # hide labels
  51. hide_conf=False, # hide confidences
  52. half=False, # use FP16 half-precision inference
  53. dnn=False, # use OpenCV DNN for ONNX inference
  54. ):
  55. source = str(source)
  56. save_img = not nosave and not source.endswith('.txt') # save inference images
  57. is_file = Path(source).suffix[1:] in (IMG_FORMATS + VID_FORMATS)
  58. is_url = source.lower().startswith(('rtsp://', 'rtmp://', 'http://', 'https://'))
  59. webcam = source.isnumeric() or source.endswith('.txt') or (is_url and not is_file)
  60. if is_url and is_file:
  61. source = check_file(source) # download
  62. # Directories
  63. save_dir = increment_path(Path(project) / name, exist_ok=exist_ok) # increment run
  64. (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
  65. # Initialize
  66. device = select_device(device)
  67. half &= device.type != 'cpu' # half precision only supported on CUDA
  68. # Load model
  69. w = str(weights[0] if isinstance(weights, list) else weights)
  70. classify, suffix, suffixes = False, Path(w).suffix.lower(), ['.pt', '.onnx', '.tflite', '.pb', '']
  71. check_suffix(w, suffixes) # check weights have acceptable suffix
  72. pt, onnx, tflite, pb, saved_model = (suffix == x for x in suffixes) # backend booleans
  73. stride, names = 64, [f'class{i}' for i in range(1000)] # assign defaults
  74. if pt:
  75. model = torch.jit.load(w) if 'torchscript' in w else attempt_load(weights, map_location=device)
  76. stride = int(model.stride.max()) # model stride
  77. names = model.module.names if hasattr(model, 'module') else model.names # get class names
  78. if half:
  79. model.half() # to FP16
  80. if classify: # second-stage classifier
  81. modelc = load_classifier(name='resnet50', n=2) # initialize
  82. modelc.load_state_dict(torch.load('resnet50.pt', map_location=device)['model']).to(device).eval()
  83. elif onnx:
  84. if dnn:
  85. check_requirements(('opencv-python>=4.5.4',))
  86. net = cv2.dnn.readNetFromONNX(w)
  87. else:
  88. check_requirements(('onnx', 'onnxruntime-gpu' if torch.has_cuda else 'onnxruntime'))
  89. import onnxruntime
  90. session = onnxruntime.InferenceSession(w, None)
  91. else: # TensorFlow models
  92. import tensorflow as tf
  93. if pb: # https://www.tensorflow.org/guide/migrate#a_graphpb_or_graphpbtxt
  94. def wrap_frozen_graph(gd, inputs, outputs):
  95. x = tf.compat.v1.wrap_function(lambda: tf.compat.v1.import_graph_def(gd, name=""), []) # wrapped import
  96. return x.prune(tf.nest.map_structure(x.graph.as_graph_element, inputs),
  97. tf.nest.map_structure(x.graph.as_graph_element, outputs))
  98. graph_def = tf.Graph().as_graph_def()
  99. graph_def.ParseFromString(open(w, 'rb').read())
  100. frozen_func = wrap_frozen_graph(gd=graph_def, inputs="x:0", outputs="Identity:0")
  101. elif saved_model:
  102. model = tf.keras.models.load_model(w)
  103. elif tflite:
  104. if "edgetpu" in w: # https://www.tensorflow.org/lite/guide/python#install_tensorflow_lite_for_python
  105. import tflite_runtime.interpreter as tflri
  106. delegate = {'Linux': 'libedgetpu.so.1', # install libedgetpu https://coral.ai/software/#edgetpu-runtime
  107. 'Darwin': 'libedgetpu.1.dylib',
  108. 'Windows': 'edgetpu.dll'}[platform.system()]
  109. interpreter = tflri.Interpreter(model_path=w, experimental_delegates=[tflri.load_delegate(delegate)])
  110. else:
  111. interpreter = tf.lite.Interpreter(model_path=w) # load TFLite model
  112. interpreter.allocate_tensors() # allocate
  113. input_details = interpreter.get_input_details() # inputs
  114. output_details = interpreter.get_output_details() # outputs
  115. int8 = input_details[0]['dtype'] == np.uint8 # is TFLite quantized uint8 model
  116. imgsz = check_img_size(imgsz, s=stride) # check image size
  117. # Dataloader
  118. if webcam:
  119. view_img = check_imshow()
  120. cudnn.benchmark = True # set True to speed up constant image size inference
  121. dataset = LoadStreams(source, img_size=imgsz, stride=stride, auto=pt)
  122. bs = len(dataset) # batch_size
  123. else:
  124. dataset = LoadImages(source, img_size=imgsz, stride=stride, auto=pt)
  125. bs = 1 # batch_size
  126. vid_path, vid_writer = [None] * bs, [None] * bs
  127. # Run inference
  128. if pt and device.type != 'cpu':
  129. model(torch.zeros(1, 3, *imgsz).to(device).type_as(next(model.parameters()))) # run once
  130. dt, seen = [0.0, 0.0, 0.0], 0
  131. for path, img, im0s, vid_cap, s in dataset:
  132. t1 = time_sync()
  133. if onnx:
  134. img = img.astype('float32')
  135. else:
  136. img = torch.from_numpy(img).to(device)
  137. img = img.half() if half else img.float() # uint8 to fp16/32
  138. img /= 255 # 0 - 255 to 0.0 - 1.0
  139. if len(img.shape) == 3:
  140. img = img[None] # expand for batch dim
  141. t2 = time_sync()
  142. dt[0] += t2 - t1
  143. # Inference
  144. if pt:
  145. visualize = increment_path(save_dir / Path(path).stem, mkdir=True) if visualize else False
  146. pred = model(img, augment=augment, visualize=visualize)[0]
  147. elif onnx:
  148. if dnn:
  149. net.setInput(img)
  150. pred = torch.tensor(net.forward())
  151. else:
  152. pred = torch.tensor(session.run([session.get_outputs()[0].name], {session.get_inputs()[0].name: img}))
  153. else: # tensorflow model (tflite, pb, saved_model)
  154. imn = img.permute(0, 2, 3, 1).cpu().numpy() # image in numpy
  155. if pb:
  156. pred = frozen_func(x=tf.constant(imn)).numpy()
  157. elif saved_model:
  158. pred = model(imn, training=False).numpy()
  159. elif tflite:
  160. if int8:
  161. scale, zero_point = input_details[0]['quantization']
  162. imn = (imn / scale + zero_point).astype(np.uint8) # de-scale
  163. interpreter.set_tensor(input_details[0]['index'], imn)
  164. interpreter.invoke()
  165. pred = interpreter.get_tensor(output_details[0]['index'])
  166. if int8:
  167. scale, zero_point = output_details[0]['quantization']
  168. pred = (pred.astype(np.float32) - zero_point) * scale # re-scale
  169. pred[..., 0] *= imgsz[1] # x
  170. pred[..., 1] *= imgsz[0] # y
  171. pred[..., 2] *= imgsz[1] # w
  172. pred[..., 3] *= imgsz[0] # h
  173. pred = torch.tensor(pred)
  174. t3 = time_sync()
  175. dt[1] += t3 - t2
  176. # NMS
  177. pred = non_max_suppression(pred, conf_thres, iou_thres, classes, agnostic_nms, max_det=max_det)
  178. dt[2] += time_sync() - t3
  179. # Second-stage classifier (optional)
  180. if classify:
  181. pred = apply_classifier(pred, modelc, img, im0s)
  182. # Process predictions
  183. for i, det in enumerate(pred): # per image
  184. seen += 1
  185. if webcam: # batch_size >= 1
  186. p, im0, frame = path[i], im0s[i].copy(), dataset.count
  187. s += f'{i}: '
  188. else:
  189. p, im0, frame = path, im0s.copy(), getattr(dataset, 'frame', 0)
  190. p = Path(p) # to Path
  191. save_path = str(save_dir / p.name) # img.jpg
  192. txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}') # img.txt
  193. s += '%gx%g ' % img.shape[2:] # print string
  194. gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
  195. imc = im0.copy() if save_crop else im0 # for save_crop
  196. annotator = Annotator(im0, line_width=line_thickness, example=str(names))
  197. if len(det):
  198. # Rescale boxes from img_size to im0 size
  199. det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
  200. # Print results
  201. for c in det[:, -1].unique():
  202. n = (det[:, -1] == c).sum() # detections per class
  203. s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
  204. # Write results
  205. for *xyxy, conf, cls in reversed(det):
  206. if save_txt: # Write to file
  207. xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
  208. line = (cls, *xywh, conf) if save_conf else (cls, *xywh) # label format
  209. with open(txt_path + '.txt', 'a') as f:
  210. f.write(('%g ' * len(line)).rstrip() % line + '\n')
  211. if save_img or save_crop or view_img: # Add bbox to image
  212. c = int(cls) # integer class
  213. label = None if hide_labels else (names[c] if hide_conf else f'{names[c]} {conf:.2f}')
  214. annotator.box_label(xyxy, label, color=colors(c, True))
  215. if save_crop:
  216. save_one_box(xyxy, imc, file=save_dir / 'crops' / names[c] / f'{p.stem}.jpg', BGR=True)
  217. # Print time (inference-only)
  218. LOGGER.info(f'{s}Done. ({t3 - t2:.3f}s)')
  219. # Stream results
  220. im0 = annotator.result()
  221. if view_img:
  222. cv2.imshow(str(p), im0)
  223. cv2.waitKey(1) # 1 millisecond
  224. # Save results (image with detections)
  225. if save_img:
  226. if dataset.mode == 'image':
  227. cv2.imwrite(save_path, im0)
  228. else: # 'video' or 'stream'
  229. if vid_path[i] != save_path: # new video
  230. vid_path[i] = save_path
  231. if isinstance(vid_writer[i], cv2.VideoWriter):
  232. vid_writer[i].release() # release previous video writer
  233. if vid_cap: # video
  234. fps = vid_cap.get(cv2.CAP_PROP_FPS)
  235. w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
  236. h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
  237. else: # stream
  238. fps, w, h = 30, im0.shape[1], im0.shape[0]
  239. save_path += '.mp4'
  240. vid_writer[i] = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*'mp4v'), fps, (w, h))
  241. vid_writer[i].write(im0)
  242. # Print results
  243. t = tuple(x / seen * 1E3 for x in dt) # speeds per image
  244. LOGGER.info(f'Speed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape {(1, 3, *imgsz)}' % t)
  245. if save_txt or save_img:
  246. s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
  247. LOGGER.info(f"Results saved to {colorstr('bold', save_dir)}{s}")
  248. if update:
  249. strip_optimizer(weights) # update model (to fix SourceChangeWarning)
  250. def parse_opt():
  251. parser = argparse.ArgumentParser()
  252. parser.add_argument('--weights', nargs='+', type=str, default=ROOT / 'yolov5s.pt', help='model path(s)')
  253. parser.add_argument('--source', type=str, default=ROOT / 'data/images', help='file/dir/URL/glob, 0 for webcam')
  254. parser.add_argument('--imgsz', '--img', '--img-size', nargs='+', type=int, default=[640], help='inference size h,w')
  255. parser.add_argument('--conf-thres', type=float, default=0.25, help='confidence threshold')
  256. parser.add_argument('--iou-thres', type=float, default=0.45, help='NMS IoU threshold')
  257. parser.add_argument('--max-det', type=int, default=1000, help='maximum detections per image')
  258. parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
  259. parser.add_argument('--view-img', action='store_true', help='show results')
  260. parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')
  261. parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')
  262. parser.add_argument('--save-crop', action='store_true', help='save cropped prediction boxes')
  263. parser.add_argument('--nosave', action='store_true', help='do not save images/videos')
  264. parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --classes 0, or --classes 0 2 3')
  265. parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')
  266. parser.add_argument('--augment', action='store_true', help='augmented inference')
  267. parser.add_argument('--visualize', action='store_true', help='visualize features')
  268. parser.add_argument('--update', action='store_true', help='update all models')
  269. parser.add_argument('--project', default=ROOT / 'runs/detect', help='save results to project/name')
  270. parser.add_argument('--name', default='exp', help='save results to project/name')
  271. parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
  272. parser.add_argument('--line-thickness', default=3, type=int, help='bounding box thickness (pixels)')
  273. parser.add_argument('--hide-labels', default=False, action='store_true', help='hide labels')
  274. parser.add_argument('--hide-conf', default=False, action='store_true', help='hide confidences')
  275. parser.add_argument('--half', action='store_true', help='use FP16 half-precision inference')
  276. parser.add_argument('--dnn', action='store_true', help='use OpenCV DNN for ONNX inference')
  277. opt = parser.parse_args()
  278. opt.imgsz *= 2 if len(opt.imgsz) == 1 else 1 # expand
  279. print_args(FILE.stem, opt)
  280. return opt
  281. def main(opt):
  282. check_requirements(exclude=('tensorboard', 'thop'))
  283. run(**vars(opt))
  284. if __name__ == "__main__":
  285. opt = parse_opt()
  286. main(opt)