Skip to content

Commit e6490b1

Browse files
JackWoo0831JackWoo0831
authored
Update README.md
1 parent e47add3 commit e6490b1

File tree

1 file changed

+1
-176
lines changed

1 file changed

+1
-176
lines changed

‎README.md

Lines changed: 1 addition & 176 deletions
Original file line numberDiff line numberDiff line change
@@ -1,176 +1 @@
1-
# Official YOLOv7
2-
3-
Implementation of paper - [YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors](https://arxiv.org/abs/2207.02696)
4-
5-
<div align="center">
6-
<a href="./">
7-
<img src="./figure/performance.png" width="79%"/>
8-
</a>
9-
</div>
10-
11-
## Web Demo
12-
13-
- Integrated into [Huggingface Spaces 🤗](https://huggingface.co/spaces/akhaliq/yolov7) using [Gradio](https://github.com/gradio-app/gradio). Try out the Web Demo [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/akhaliq/yolov7)
14-
15-
## Performance
16-
17-
MS COCO
18-
19-
| Model | Test Size | AP<sup>test</sup> | AP<sub>50</sub><sup>test</sup> | AP<sub>75</sub><sup>test</sup> | batch 1 fps | batch 32 average time |
20-
| :-- | :-: | :-: | :-: | :-: | :-: | :-: |
21-
| [**YOLOv7**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt) | 640 | **51.4%** | **69.7%** | **55.9%** | 161 *fps* | 2.8 *ms* |
22-
| [**YOLOv7-X**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7x.pt) | 640 | **53.1%** | **71.2%** | **57.8%** | 114 *fps* | 4.3 *ms* |
23-
| | | | | | | |
24-
| [**YOLOv7-W6**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-w6.pt) | 1280 | **54.9%** | **72.6%** | **60.1%** | 84 *fps* | 7.6 *ms* |
25-
| [**YOLOv7-E6**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6.pt) | 1280 | **56.0%** | **73.5%** | **61.2%** | 56 *fps* | 12.3 *ms* |
26-
| [**YOLOv7-D6**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-d6.pt) | 1280 | **56.6%** | **74.0%** | **61.8%** | 44 *fps* | 15.0 *ms* |
27-
| [**YOLOv7-E6E**](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6e.pt) | 1280 | **56.8%** | **74.4%** | **62.1%** | 36 *fps* | 18.7 *ms* |
28-
29-
## Installation
30-
31-
Docker environment (recommended)
32-
<details><summary> <b>Expand</b> </summary>
33-
34-
``` bash
35-
# create the docker container, you can change the share memory size if you have more.
36-
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
37-
38-
# apt install required packages
39-
apt update
40-
apt install -y zip htop screen libgl1-mesa-glx
41-
42-
# pip install required packages
43-
pip install seaborn thop
44-
45-
# go to code folder
46-
cd /yolov7
47-
```
48-
49-
</details>
50-
51-
## Testing
52-
53-
[`yolov7.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt) [`yolov7x.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7x.pt) [`yolov7-w6.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-w6.pt) [`yolov7-e6.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6.pt) [`yolov7-d6.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-d6.pt) [`yolov7-e6e.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6e.pt)
54-
55-
``` bash
56-
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
57-
```
58-
59-
You will get the results:
60-
61-
```
62-
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.51206
63-
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.69730
64-
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.55521
65-
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.35247
66-
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.55937
67-
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.66693
68-
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.38453
69-
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.63765
70-
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.68772
71-
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.53766
72-
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.73549
73-
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.83868
74-
```
75-
76-
To measure accuracy, download [COCO-annotations for Pycocotools](http://images.cocodataset.org/annotations/annotations_trainval2017.zip).
77-
78-
## Training
79-
80-
Data preparation
81-
82-
``` bash
83-
bash scripts/get_coco.sh
84-
```
85-
86-
* Download MS COCO dataset images ([train](http://images.cocodataset.org/zips/train2017.zip), [val](http://images.cocodataset.org/zips/val2017.zip), [test](http://images.cocodataset.org/zips/test2017.zip)) and [labels](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/coco2017labels-segments.zip). 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](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/coco2017labels-segments.zip)
87-
88-
Single GPU training
89-
90-
``` bash
91-
# train p5 models
92-
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
93-
94-
# train p6 models
95-
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
96-
```
97-
98-
Multiple GPU training
99-
100-
``` bash
101-
# train p5 models
102-
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
103-
104-
# train p6 models
105-
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
106-
```
107-
108-
## Transfer learning
109-
110-
[`yolov7_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7_training.pt) [`yolov7x_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7x_training.pt) [`yolov7-w6_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-w6_training.pt) [`yolov7-e6_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6_training.pt) [`yolov7-d6_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-d6_training.pt) [`yolov7-e6e_training.pt`](https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7-e6e_training.pt)
111-
112-
Single GPU finetuning for custom dataset
113-
114-
``` bash
115-
# finetune p5 models
116-
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
117-
118-
# finetune p6 models
119-
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
120-
```
121-
122-
## Re-parameterization
123-
124-
See [reparameterization.ipynb](tools/reparameterization.ipynb)
125-
126-
## Inference
127-
128-
``` bash
129-
python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source inference/images/horses.jpg
130-
```
131-
132-
<div align="center">
133-
<a href="./">
134-
<img src="./figure/horses_prediction.jpg" width="59%"/>
135-
</a>
136-
</div>
137-
138-
## Citation
139-
140-
```
141-
@article{wang2022yolov7,
142-
title={{YOLOv7}: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors},
143-
author={Wang, Chien-Yao and Bochkovskiy, Alexey and Liao, Hong-Yuan Mark},
144-
journal={arXiv preprint arXiv:2207.02696},
145-
year={2022}
146-
}
147-
```
148-
149-
## Teaser
150-
151-
Yolov7-mask & YOLOv7-pose
152-
153-
<div align="center">
154-
<a href="./">
155-
<img src="./figure/mask.png" width="56%"/>
156-
</a>
157-
<a href="./">
158-
<img src="./figure/pose.png" width="42%"/>
159-
</a>
160-
</div>
161-
162-
## Acknowledgements
163-
164-
<details><summary> <b>Expand</b> </summary>
165-
166-
* [https://github.com/AlexeyAB/darknet](https://github.com/AlexeyAB/darknet)
167-
* [https://github.com/WongKinYiu/yolor](https://github.com/WongKinYiu/yolor)
168-
* [https://github.com/WongKinYiu/PyTorch_YOLOv4](https://github.com/WongKinYiu/PyTorch_YOLOv4)
169-
* [https://github.com/WongKinYiu/ScaledYOLOv4](https://github.com/WongKinYiu/ScaledYOLOv4)
170-
* [https://github.com/Megvii-BaseDetection/YOLOX](https://github.com/Megvii-BaseDetection/YOLOX)
171-
* [https://github.com/ultralytics/yolov3](https://github.com/ultralytics/yolov3)
172-
* [https://github.com/ultralytics/yolov5](https://github.com/ultralytics/yolov5)
173-
* [https://github.com/DingXiaoH/RepVGG](https://github.com/DingXiaoH/RepVGG)
174-
* [https://github.com/JUGGHM/OREPA_CVPR2022](https://github.com/JUGGHM/OREPA_CVPR2022)
175-
176-
</details>
1+

0 commit comments

Comments
 (0)