Custom YOLOv7

This is a customization of the official YOLOv7 implementation. The original Readme starts here. The customisations are:

• support for stacked images (or called multi-frame images)
• support for images with 4 channels
• tiling and sampling of the images
• center point distance instead of IoU
• measuring of inference

Stacked Images

The input images should have the numpy .npy format. Use the multi-frame parameter to set the number of frames that such an input image contain. This parameter is available in train.py, test.py, detect.py and train_aux.py. During training with stacked images it is ensured, that all the individual frames of a stack experience the same augmentations. For measuring the detection time of stacked images use --mode 2 in measure_inference.py.

4 channel images

This is intended for images with an additional background subtracted channel, that carries the background mask. This additional channel is excluded from pixelwise augmentations (e.g. HSV gains/losses). Use the four-channels parameter to enable training, testing or detection with such images. For measuring the detection, background subtraction and illumination smoothing time of four-channel-images use --mode 1 in measure_inference.py.

Tiling and Sampling

Use the tiles parameter to cut input images into smaller parts during training or testing. The given number determines the number of cuts that are made both vertically and horizontally. The resulting tiles are written to file in a newly created folder images_tiled in the data input folder. This folder is not removed and if the network is trained or tested again with the tiles parameter, the tiles will simply be read from the folder. If you adjust the number of tiles be sure to delete the old images_tiled folder.

Use the no-class parameter to randomly undersample images with no class present. The given number determines the percentage of images without classes in the input data. If the input data already contains fewer images without classes no undersampling is conducted. Just like for the tiling the sampled input images are written to a new folder images_filtered, which will be reused if present.

If both the tiles and no-class parameter are present the undersampling will be applied on the tiled images.

Center Point Distance

Use the center-point parameter to indicate that correct predictions are not determined through a minimum IoU with the ground truth, but through a maximum distance of the centers of bounding boxes. The distance is set between 5 and 20 pixels by default. The calculated [email protected] equates to a maximal distance of 20 pixels and map@[.5:.95] equates to a distance between 20 and 5 pixels.

Measure time

Use the measure-inference.py script to measure detection time and also background subtraction and illumination smoothing time.

Official YOLOv7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors

Web Demo

• Integrated into Huggingface Spaces 🤗 using Gradio. Try out the Web Demo

Performance

MS COCO

Model	Test Size	APtest	AP50test	AP75test	batch 1 fps	batch 32 average time
YOLOv7	640	51.4%	69.7%	55.9%	161 fps	2.8 ms
YOLOv7-X	640	53.1%	71.2%	57.8%	114 fps	4.3 ms
YOLOv7-W6	1280	54.9%	72.6%	60.1%	84 fps	7.6 ms
YOLOv7-E6	1280	56.0%	73.5%	61.2%	56 fps	12.3 ms
YOLOv7-D6	1280	56.6%	74.0%	61.8%	44 fps	15.0 ms
YOLOv7-E6E	1280	56.8%	74.4%	62.1%	36 fps	18.7 ms

Installation

Docker environment (recommended)

Expand

# create the docker container, you can change the share memory size if you have more.
nvidia-docker run --name yolov7 -it -v your_coco_path/:/coco/ -v your_code_path/:/yolov7 --shm-size=64g nvcr.io/nvidia/pytorch:21.08-py3

# apt install required packages
apt update
apt install -y zip htop screen libgl1-mesa-glx

# pip install required packages
pip install seaborn thop

# go to code folder
cd /yolov7

Testing

yolov7.pt yolov7x.pt yolov7-w6.pt yolov7-e6.pt yolov7-d6.pt yolov7-e6e.pt

python test.py --data data/coco.yaml --img 640 --batch 32 --conf 0.001 --iou 0.65 --device 0 --weights yolov7.pt --name yolov7_640_val

You will get the results:

 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.51206
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.69730
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.55521
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.35247
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.55937
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.66693
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.38453
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.63765
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.68772
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.53766
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.73549
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.83868

To measure accuracy, download COCO-annotations for Pycocotools to the ./coco/annotations/instances_val2017.json

Training

Data preparation

bash scripts/get_coco.sh

• Download MS COCO dataset images (train, val, test) and labels. If you have previously used a different version of YOLO, we strongly recommend that you delete train2017.cache and val2017.cache files, and redownload labels

Single GPU training

# train p5 models
python train.py --workers 8 --device 0 --batch-size 32 --data data/coco.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7 --hyp data/hyp.scratch.p5.yaml

# train p6 models
python train_aux.py --workers 8 --device 0 --batch-size 16 --data data/coco.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6.yaml --weights '' --name yolov7-w6 --hyp data/hyp.scratch.p6.yaml

Multiple GPU training

# train p5 models
python -m torch.distributed.launch --nproc_per_node 4 --master_port 9527 train.py --workers 8 --device 0,1,2,3 --sync-bn --batch-size 128 --data data/coco.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7 --hyp data/hyp.scratch.p5.yaml

# train p6 models
python -m torch.distributed.launch --nproc_per_node 8 --master_port 9527 train_aux.py --workers 8 --device 0,1,2,3,4,5,6,7 --sync-bn --batch-size 128 --data data/coco.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6.yaml --weights '' --name yolov7-w6 --hyp data/hyp.scratch.p6.yaml

Transfer learning

yolov7_training.pt yolov7x_training.pt yolov7-w6_training.pt yolov7-e6_training.pt yolov7-d6_training.pt yolov7-e6e_training.pt

Single GPU finetuning for custom dataset

# finetune p5 models
python train.py --workers 8 --device 0 --batch-size 32 --data data/custom.yaml --img 640 640 --cfg cfg/training/yolov7-custom.yaml --weights 'yolov7_training.pt' --name yolov7-custom --hyp data/hyp.scratch.custom.yaml

# finetune p6 models
python train_aux.py --workers 8 --device 0 --batch-size 16 --data data/custom.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6-custom.yaml --weights 'yolov7-w6_training.pt' --name yolov7-w6-custom --hyp data/hyp.scratch.custom.yaml

Re-parameterization

See reparameterization.ipynb

Inference

On video:

python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source yourvideo.mp4

On image:

python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source inference/images/horses.jpg

Export

Pytorch to CoreML (and inference on MacOS/iOS)

Pytorch to ONNX with NMS (and inference)

python export.py --weights yolov7-tiny.pt --grid --end2end --simplify \
        --topk-all 100 --iou-thres 0.65 --conf-thres 0.35 --img-size 640 640 --max-wh 640

Pytorch to TensorRT with NMS (and inference)

wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-tiny.pt
python export.py --weights ./yolov7-tiny.pt --grid --end2end --simplify --topk-all 100 --iou-thres 0.65 --conf-thres 0.35 --img-size 640 640
git clone https://github.com/Linaom1214/tensorrt-python.git
python ./tensorrt-python/export.py -o yolov7-tiny.onnx -e yolov7-tiny-nms.trt -p fp16

Pytorch to TensorRT another way

Expand

wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-tiny.pt
python export.py --weights yolov7-tiny.pt --grid --include-nms
git clone https://github.com/Linaom1214/tensorrt-python.git
python ./tensorrt-python/export.py -o yolov7-tiny.onnx -e yolov7-tiny-nms.trt -p fp16

# Or use trtexec to convert ONNX to TensorRT engine
/usr/src/tensorrt/bin/trtexec --onnx=yolov7-tiny.onnx --saveEngine=yolov7-tiny-nms.trt --fp16

Tested with: Python 3.7.13, Pytorch 1.12.0+cu113

Pose estimation

code yolov7-w6-pose.pt

See keypoint.ipynb.

Instance segmentation

code yolov7-mask.pt

See instance.ipynb.

Citation

@article{wang2022yolov7,
  title={{YOLOv7}: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors},
  author={Wang, Chien-Yao and Bochkovskiy, Alexey and Liao, Hong-Yuan Mark},
  journal={arXiv preprint arXiv:2207.02696},
  year={2022}
}

Teaser

Yolov7-semantic & YOLOv7-panoptic & YOLOv7-caption

Acknowledgements

Expand

• https://github.com/AlexeyAB/darknet
• https://github.com/WongKinYiu/yolor
• https://github.com/WongKinYiu/PyTorch_YOLOv4
• https://github.com/WongKinYiu/ScaledYOLOv4
• https://github.com/Megvii-BaseDetection/YOLOX
• https://github.com/ultralytics/yolov3
• https://github.com/ultralytics/yolov5
• https://github.com/DingXiaoH/RepVGG
• https://github.com/JUGGHM/OREPA_CVPR2022
• https://github.com/TexasInstruments/edgeai-yolov5/tree/yolo-pose