Now it can count objects with the command "python tools/infer.py --weights yolov6s.pt --class 56 --source camera".

README.md

@@ -136,3 +136,6 @@ Your can also specify a checkpoint path to `--resume` parameter by
  * [YOLOv6 web demo](https://huggingface.co/spaces/nateraw/yolov6) on [Huggingface Spaces](https://huggingface.co/spaces) with [Gradio](https://github.com/gradio-app/gradio). [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/nateraw/yolov6)
  * Tutorial: [How to train YOLOv6 on a custom dataset](https://blog.roboflow.com/how-to-train-yolov6-on-a-custom-dataset/) <a href="https://colab.research.google.com/drive/1YnbqOinBZV-c9I7fk_UL6acgnnmkXDMM"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a>
  * Demo of YOLOv6 inference on Google Colab [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/mahdilamb/YOLOv6/blob/main/inference.ipynb)
+
+
+Now it can count objects with the command "python tools/infer.py --weights yolov6s.pt --class 56 --source camera".

configs/repopt/yolov6s_hs.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],
@@ -51,4 +51,4 @@
 )
 
 # Choose Rep-block by the training Mode, choices=["repvgg", "hyper-search", "repopt"]
-training_mode='hyper_search'
+training_mode = 'hyper_search'

configs/repopt/yolov6s_opt.py

@@ -9,12 +9,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],
@@ -52,4 +52,4 @@
 )
 
 # Choose Rep-block by the training Mode, choices=["repvgg", "hyper-search", "repopt"]
-training_mode='repopt'
+training_mode = 'repopt'

configs/yolov6_tiny.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

configs/yolov6_tiny_finetune.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

configs/yolov6n.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

configs/yolov6n_finetune.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

configs/yolov6s.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

configs/yolov6s_finetune.py

@@ -8,12 +8,12 @@
         type='EfficientRep',
         num_repeats=[1, 6, 12, 18, 6],
         out_channels=[64, 128, 256, 512, 1024],
-        ),
+    ),
     neck=dict(
         type='RepPAN',
         num_repeats=[12, 12, 12, 12],
         out_channels=[256, 128, 128, 256, 256, 512],
-        ),
+    ),
     head=dict(
         type='EffiDeHead',
         in_channels=[128, 256, 512],

data/coco.yaml

@@ -1,7 +1,7 @@
 # COCO 2017 dataset http://cocodataset.org
-train: ../coco/images/train2017 # 118287 images
-val: ../coco/images/val2017  # 5000 images
-test: ../coco/images/test2017
+train: C:/Users/ahmtb/YOLOv6coco/images/train2017 # 118287 images
+val: C:/Users/ahmtb/YOLOv6coco/images/train2017  # 5000 images
+test: C:/Users/ahmtb/YOLOv6coco/images/train2017
 anno_path: ../coco/annotations/instances_val2017.json
 # number of classes
 nc: 80
@@ -17,4 +17,4 @@ names: [ 'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', '
          'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch',
          'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone',
          'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear',
-         'hair drier', 'toothbrush' ]
+         'hair drier', 'toothbrush' ]

data/dataset.yaml

@@ -7,5 +7,5 @@ test: ../custom_dataset/images/test # test images (optional)
 is_coco: False
 # Classes
 nc: 20  # number of classes
-names: ['aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
-        'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor']  # class names
+names: [ 'aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
+         'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor' ]  # class names

data/voc.yaml

@@ -7,5 +7,5 @@ test: VOCdevkit/voc_07_12/images/val # test images (optional)
 is_coco: False
 # Classes
 nc: 20  # number of classes
-names: ['aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
-        'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor']  # class names
+names: [ 'aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
+         'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor' ]  # class names

deploy/ONNX/README.md

@@ -1,20 +1,20 @@
 # Export ONNX Model
 
 ## Check requirements
+
 ```shell
 pip install onnx>=1.10.0
 ```
 
 ## Export script
+
 ```shell
 python ./deploy/ONNX/export_onnx.py \
     --weights yolov6s.pt \
     --img 640 \
     --batch 1
 ```
 
-
-
 #### Description of all arguments
 
 - `--weights` : The path of yolov6 model weights.
@@ -45,6 +45,7 @@ Now YOLOv6 supports end to end detect for onnxruntime and TensorRT !
 If you want to deploy in TensorRT, make sure you have installed TensorRT >= 8.0.0 !
 
 ### onnxruntime backend
+
 #### Usage
 
 ```bash
@@ -77,7 +78,8 @@ python ./deploy/ONNX/export_onnx.py \
     --trt-version 7
 ```
 
-You will get an onnx with **[BatchedNMSDynamic_TRT](https://github.com/triple-Mu/TensorRT/tree/main/plugin/batchedNMSPlugin)** plugin .
+You will get an onnx
+with **[BatchedNMSDynamic_TRT](https://github.com/triple-Mu/TensorRT/tree/main/plugin/batchedNMSPlugin)** plugin .
 
 ### TensorRT backend (TensorRT version>= 8.0.0)
 
@@ -92,7 +94,8 @@ python ./deploy/ONNX/export_onnx.py \
     --trt-version 8
 ```
 
-You will get an onnx with **[EfficientNMS_TRT](https://github.com/NVIDIA/TensorRT/tree/main/plugin/efficientNMSPlugin)** plugin .
+You will get an onnx with **[EfficientNMS_TRT](https://github.com/NVIDIA/TensorRT/tree/main/plugin/efficientNMSPlugin)**
+plugin .
 
 ### Outputs Description
 
@@ -108,8 +111,9 @@ The onnx outputs are as shown :
 
 ```det_classes``` means the category of every topk(100) objects .
 
+You can export TensorRT engine
+use [trtexec](https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#trtexec-ovr) tools.
 
-You can export TensorRT engine use [trtexec](https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#trtexec-ovr) tools.
 #### Usage
 
 For both TensorRT-7 and TensorRT-8  `trtexec`  tool is avaiable.
@@ -124,6 +128,7 @@ trtexec --onnx=yolov6s.onnx \
 ## Evaluate TensorRT model's performance
 
 When we get the TensorRT model, we can evalute its performance by:
+
 ```
 python deploy/ONNX/eval_trt.py --weights yolov6s.engine --batch-size=1 --data data/coco.yaml
 ```

deploy/ONNX/eval_trt.py

@@ -38,14 +38,14 @@ def run(data,
         task='val',
         device='',
         save_dir='',
-        name = ''
+        name=''
         ):
     """
     TensorRT models's evaluation process.
     """
 
-     # task
-    assert task== 'val', f'task type can only be val, however you set it to {task}'
+    # task
+    assert task == 'val', f'task type can only be val, however you set it to {task}'
 
     save_dir = str(increment_name(osp.join(save_dir, name)))
     os.makedirs(save_dir, exist_ok=True)
@@ -57,9 +57,9 @@ def run(data,
 
     # init
     val = Evaler(data, batch_size, img_size, None, \
-                None, device, False, save_dir)
+                 None, device, False, save_dir)
 
-    dataloader,pred_result = val.eval_trt(weights)
+    dataloader, pred_result = val.eval_trt(weights)
     eval_result = val.eval_model(pred_result, dummy_model, dataloader, task)
     return eval_result
 

deploy/ONNX/export_onnx.py

@@ -19,7 +19,6 @@
 from yolov6.utils.checkpoint import load_checkpoint
 from io import BytesIO
 
-
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--weights', type=str, default='./yolov6s.pt', help='weights path')
@@ -31,7 +30,8 @@
     parser.add_argument('--end2end', action='store_true', help='export end2end onnx')
     parser.add_argument('--trt-version', type=int, default=8, help='tensorrt version')
     parser.add_argument('--with-preprocess', action='store_true', help='export bgr2rgb and normalize')
-    parser.add_argument('--max-wh', type=int, default=None, help='None for tensorrt nms, int value for onnx-runtime nms')
+    parser.add_argument('--max-wh', type=int, default=None,
+                        help='None for tensorrt nms, int value for onnx-runtime nms')
     parser.add_argument('--topk-all', type=int, default=100, help='topk objects for every images')
     parser.add_argument('--iou-thres', type=float, default=0.45, help='iou threshold for NMS')
     parser.add_argument('--conf-thres', type=float, default=0.25, help='conf threshold for NMS')
@@ -66,8 +66,10 @@
             m.inplace = args.inplace
     if args.end2end:
         from yolov6.models.end2end import End2End
-        model = End2End(model, max_obj=args.topk_all, iou_thres=args.iou_thres,score_thres=args.conf_thres,
-                        max_wh=args.max_wh, device=device, trt_version=args.trt_version, with_preprocess=args.with_preprocess)
+
+        model = End2End(model, max_obj=args.topk_all, iou_thres=args.iou_thres, score_thres=args.conf_thres,
+                        max_wh=args.max_wh, device=device, trt_version=args.trt_version,
+                        with_preprocess=args.with_preprocess)
 
     y = model(img)  # dry run
 
@@ -81,7 +83,7 @@
                               do_constant_folding=True,
                               input_names=['images'],
                               output_names=['num_dets', 'det_boxes', 'det_scores', 'det_classes']
-                              if args.end2end and args.max_wh is None else ['outputs'],)
+                              if args.end2end and args.max_wh is None else ['outputs'], )
             f.seek(0)
             # Checks
             onnx_model = onnx.load(f)  # load onnx model
@@ -96,6 +98,7 @@
         if args.simplify:
             try:
                 import onnxsim
+
                 LOGGER.info('\nStarting to simplify ONNX...')
                 onnx_model, check = onnxsim.simplify(onnx_model)
                 assert check, 'assert check failed'
@@ -111,4 +114,4 @@
     if args.end2end:
         if args.max_wh is None:
             LOGGER.info('\nYou can export tensorrt engine use trtexec tools.\nCommand is:')
-            LOGGER.info(f'trtexec --onnx={export_file} --saveEngine={export_file.replace(".onnx",".engine")}')
+            LOGGER.info(f'trtexec --onnx={export_file} --saveEngine={export_file.replace(".onnx", ".engine")}')

deploy/OpenVINO/README.md

@@ -1,23 +1,27 @@
 ## Export OpenVINO Model
 
 ### Check requirements
+
 ```shell
 pip install --upgrade pip
 pip install openvino-dev
 ```
 
 ### Export script
+
 ```shell
 python deploy/OpenVINO/export_openvino.py --weights yolov6s.pt --img 640 --batch 1
 
 ```
 
 ### Download
+
 * [YOLOv6-nano](https://github.com/meituan/YOLOv6/releases/download/0.1.0/yolov6n_openvino.tar.gz)
 * [YOLOv6-tiny](https://github.com/meituan/YOLOv6/releases/download/0.1.0/yolov6n_openvino.tar.gz)
 * [YOLOv6-s](https://github.com/meituan/YOLOv6/releases/download/0.1.0/yolov6n_openvino.tar.gz)
 
 ### Speed test
+
 ```shell
 benchmark_app -m yolov6s_openvino/yolov6s.xml -i data/images/image1.jpg -d CPU -niter 100 -progress
 

deploy/OpenVINO/export_openvino.py

@@ -19,7 +19,6 @@
 from yolov6.utils.events import LOGGER
 from yolov6.utils.checkpoint import load_checkpoint
 
-
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--weights', type=str, default='./yolov6s.pt', help='weights path')
@@ -68,7 +67,7 @@
                           do_constant_folding=True,
                           input_names=['image_arrays'],
                           output_names=['outputs'],
-                         )
+                          )
 
         # Checks
         onnx_model = onnx.load(export_file)  # load onnx model

docs/About_naming_yolov6.md

@@ -1,7 +1,12 @@
 # About the naming of YOLOv6
 
 ### WHY named YOLOv6 ?
-The full name is actually MT-YOLOv6, which is called YOLOv6 for brevity. Our work is majorly inspired by the original idea of the one-stage YOLO detection algorithm and the implementation has leveraged various techniques and tricks of former relevant work . Therefore, we named the project YOLOv6 to pay tribute to the work of YOLO series. Furthermore, we have indeed adopted some novel method and made solid engineering improvements to dedicate the algorithm to industrial applications.
+
+The full name is actually MT-YOLOv6, which is called YOLOv6 for brevity. Our work is majorly inspired by the original
+idea of the one-stage YOLO detection algorithm and the implementation has leveraged various techniques and tricks of
+former relevant work . Therefore, we named the project YOLOv6 to pay tribute to the work of YOLO series. Furthermore, we
+have indeed adopted some novel method and made solid engineering improvements to dedicate the algorithm to industrial
+applications.
 As for the project, we'll continue to improve and maintain it, contributing more values for industrial applications.
 
 P.S. We are contacting the authors of YOLO series about the naming of YOLOv6.

docs/Test_speed.md

@@ -1,6 +1,7 @@
 # Test speed
 
-This guidence explains how to reproduce speed results of YOLOv6. For fair comparison, the speed results do not contain the time cost of data pre-processing and NMS post-processing.
+This guidence explains how to reproduce speed results of YOLOv6. For fair comparison, the speed results do not contain
+the time cost of data pre-processing and NMS post-processing.
 
 ## 0. Prepare model