NetsPresso tutorial for YOLOv7 compression

January 24, 2025 · View on GitHub

NetsPresso tutorial for YOLOv7 compression

Order of the tutorial

0. Sign up
1. Install
2. Training
3. Transfer Training
4. Convert YOLOv7 to yolov7_fx.pt 1
5. Model compression with NetsPresso Python Package
6. Restore the compressed model to the original model structure
7. Retrain the compressed model
8. NetsPresso Re-parameterization
9. Convert YOLOv7 to yolov7_fx.pt 2

To get started with the NetsPresso Python package, you will need to sign up at NetsPresso.

1. Install

Clone repo and install requirements.txt in a Python>=3.7.0 environment, including PyTorch >= 1.11, < 2.0.

git clone https://github.com/Nota-NetsPresso/ModelZoo-YOLOv7.git  # clone
cd ModelZoo-YOLOv7
pip install -r requirements.txt

2. 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

3. 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

4. Convert YOLOv7 to yolov7_fx.pt 1

python export_netspresso.py --weights yolov7_training.pt --data data/coco.yaml

Executing this code will create 'yolov7_fx.pt'.

5. Model compression with NetsPresso Python Package

Upload & compress your 'yolov7_fx.pt' by using NetsPresso Python Package

5_1. Install NetsPresso Python Package

pip install netspresso

5_2. Upload & compress

First, import the packages and set a NetsPresso username and password.

from netspresso import NetsPresso

netspresso = NetsPresso(email="YOUR_EMAIL", password="YOUR_PASSWORD")

Second, upload 'model_to_compress.pt', which is the model converted to torchfx in step 4, with the following code.

# 1. Declare compressor
compressor = netspresso.compressor_v2()

# 2. Run automatic compression
compression_result = compressor.automatic_compression(
    input_shapes=[{"batch": 1, "channel": 3, "dimension": [640, 640]}],
    input_model_path="./yolov7_fx.pt",
    output_dir="./yolov7_L206.pt",
    compression_ratio=0.5,
)

More commands can be found in the official NetsPresso Python Package docs: https://nota-netspresso.github.io/PyNetsPresso-docs
Alternatively, you can do the same as above through the GUI on our website: https://console.netspresso.ai/models

6. Restore the compressed model to the original model structure

The compressed model is restored to the original model structure through the following code. This will create a fx2p_complete.pt file.

python yolov7_fx2p.py --original yolov7_training.pt --compressed yolov7_L206.pt --detect 105

7. Retrain the compressed model

The compressed model is restored to the original model structure through the following code.

python train.py --netspresso --workers 8 --device 0 --batch-size 32 --data data/coco.yaml --img 640 640 --weights fx2p_complete.pt --name yolov7 --hyp data/hyp.scratch.p5.yaml

8. NetsPresso Re-parameterization

See netspresso_reparameterization.ipynb

9. Convert YOLOv7 to yolov7_fx.pt 2

If you want to compress the compressed model?

python export_netspresso.py --netspresso --weights fx2p_complete.pt --data data/coco.yaml

Start with the following code and repeat steps 5, 6, and 7!
Now you can use the compressed model however you like!

Join our Discussion Forum for providing feedback or sharing your use cases, and if you want to talk more with Nota, please contact us here.
Or you can also do it via email(contact@nota.ai) or phone(+82 2-555-8659)!

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	AP^test	AP₅₀^test	AP₇₅^test	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

Model	Test Size	AP^box	AP₅₀^box	AP₇₅^box	AP^mask	AP₅₀^mask	AP₇₅^mask
YOLOv7-seg	640	51.4%	69.4%	55.8%	41.5%	65.5%	43.7%

Anchor free detection head

code yolov7-u6.pt

YOLOv7 with decoupled TAL head (YOLOR + YOLOv5 + YOLOv6)

Model	Test Size	AP^val	AP₅₀^val	AP₇₅^val
YOLOv7-u6	640	52.6%	69.7%	57.3%

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}
}

@article{wang2022designing,
  title={Designing Network Design Strategies Through Gradient Path Analysis},
  author={Wang, Chien-Yao and Liao, Hong-Yuan Mark and Yeh, I-Hau},
  journal={arXiv preprint arXiv:2211.04800},
  year={2022}
}