Skip to content

Commit

Permalink
update
Browse files Browse the repository at this point in the history
  • Loading branch information
@Chilicyy
Chilicyy committed Aug 28, 2023
1 parent 417c2ad commit 8bb4d26
Show file tree
Hide file tree
Showing 2 changed files with 88 additions and 315 deletions.
190 changes: 37 additions & 153 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,149 +4,67 @@

English | [简体中文](README_cn.md)

<br>

<div>
</a>
<a href="https://colab.research.google.com/github/meituan/YOLOv6/blob/main/turtorial.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a>
<a href="https://www.kaggle.com/code/housanduo/yolov6"><img src="https://kaggle.com/static/images/open-in-kaggle.svg" alt="Open In Kaggle"></a>
</div>
<br>
cp
## YOLOv6

Implementation of paper:
- [YOLOv6 v3.0: A Full-Scale Reloading](https://arxiv.org/abs/2301.05586) 🔥
- [YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications](https://arxiv.org/abs/2209.02976)

<p align="center">
<img src="assets/speed_comparision_v3.png" align="middle" width = "1000" />
</p>


## What's New
- [2023.04.28] Release [YOLOv6Lite](configs/yolov6_lite/README.md) models on mobile or CPU. ⭐️ [Mobile Benchmark](#Mobile-Benchmark)
- [2023.03.10] Release [YOLOv6-Face](https://github.com/meituan/YOLOv6/tree/yolov6-face). 🔥 [Performance](https://github.com/meituan/YOLOv6/tree/yolov6-face#performance-on-widerface)
- [2023.03.02] Update [base models](configs/base/README.md) to version 3.0.
- [2023.01.06] Release P6 models and enhance the performance of P5 models. ⭐️ [Benchmark](#Benchmark)
- [2022.11.04] Release [base models](configs/base/README.md) to simplify the training and deployment process.
- [2022.09.06] Customized quantization methods. 🚀 [Quantization Tutorial](./tools/qat/README.md)
- [2022.09.05] Release M/L models and update N/T/S models with enhanced performance.
- [2022.06.23] Release N/T/S models with excellent performance.

## Benchmark
| Model | Size | mAP<sup>val<br/>0.5:0.95 | Speed<sup>T4<br/>trt fp16 b1 <br/>(fps) | Speed<sup>T4<br/>trt fp16 b32 <br/>(fps) | Params<br/><sup> (M) | FLOPs<br/><sup> (G) |
| :----------------------------------------------------------- | ---- | :----------------------- | --------------------------------------- | ---------------------------------------- | -------------------- | ------------------- |
| [**YOLOv6-N**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6n.pt) | 640 | 37.5 | 779 | 1187 | 4.7 | 11.4 |
| [**YOLOv6-S**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6s.pt) | 640 | 45.0 | 339 | 484 | 18.5 | 45.3 |
| [**YOLOv6-M**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6m.pt) | 640 | 50.0 | 175 | 226 | 34.9 | 85.8 |
| [**YOLOv6-L**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6l.pt) | 640 | 52.8 | 98 | 116 | 59.6 | 150.7 |
| | | | | |
| [**YOLOv6-N6**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6n6.pt) | 1280 | 44.9 | 228 | 281 | 10.4 | 49.8 |
| [**YOLOv6-S6**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6s6.pt) | 1280 | 50.3 | 98 | 108 | 41.4 | 198.0 |
| [**YOLOv6-M6**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6m6.pt) | 1280 | 55.2 | 47 | 55 | 79.6 | 379.5 |
| [**YOLOv6-L6**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6l6.pt) | 1280 | 57.2 | 26 | 29 | 140.4 | 673.4 |
<details>
<summary>Table Notes</summary>

- All checkpoints are trained with self-distillation except for YOLOv6-N6/S6 models trained to 300 epochs without distillation.
- Results of the mAP and speed are evaluated on [COCO val2017](https://cocodataset.org/#download) dataset with the input resolution of 640×640 for P5 models and 1280x1280 for P6 models.
- Speed is tested with TensorRT 7.2 on T4.
## YOLOv6-Segmentation

Implementation of Instance Segmentation based on [YOLOv6 v4.0 code](https://github.com/meituan/YOLOv6/tree/main).

New Feature
- Designe two types of segment heads referring to YOLACT and SOLO

## Performance on MSCOCO
| Model | Size | mAP<sup>box<br/>50-95 | mAP<sup>mask<br/>50-95 | Speed<sup>T4<br/>trt fp16 b1 <br/>(fps) |
| :----------------------------------------------------------- | ---- | :-------------------- | ---------------------- | --------------------------------------- |
| [**YOLOv6-N-seg**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6n_seg.pt) | 640 | 35.3 | 31.2 | 645 |
| [**YOLOv6-S-seg**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6s_seg.pt) | 640 | 44.0 | 38.0 | 292 |
| [**YOLOv6-M-seg**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6m_seg.pt) | 640 | 48.2 | 41.3 | 148 |
| [**YOLOv6-L-seg**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6l_seg.pt) | 640 | 51.1 | 43.7 | 93 |
| [**YOLOv6-X-seg**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6x_seg.pt) | 640 | 52.2 | 44.8 | 47 |
| | | | | |
| [**YOLOv6-N-solo**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6n_solo.pt) | 640 | 35.7 | 31.3 | 506 |
| [**YOLOv6-S-solo**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6s_solo.pt) | 640 | 44.2 | 39.0 | 243 |
| [**YOLOv6-M-solo**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6m_solo.pt) | 640 | 48.3 | 42.2 | 120 |
| [**YOLOv6-L-solo**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6l_solo.pt) | 640 | 50.9 | 44.4 | 75 |
| [**YOLOv6-X-solo**](https://github.com/meituan/YOLOv6/releases/download/0.4.1/yolov6x_solo.pt) | 640 | 52.2 | 45.0 | 41 |

#### Table Notes
- YOLOv6-x-seg models apply YOLACT segment head, and YOLOv6-x-solo models apply SOLO segment head.
- All checkpoints are trained from scratch on COCO for 300 epochs without distillation.
- Results of the mAP and speed are evaluated on [COCO val2017](https://cocodataset.org/#download) dataset with the input resolution of 640×640.
- Speed is tested with TensorRT 8.5 on T4 without post-processing.
- Refer to [Test speed](./docs/Test_speed.md) tutorial to reproduce the speed results of YOLOv6.
- Params and FLOPs of YOLOv6 are estimated on deployed models.
</details>

<details>
<summary>Legacy models</summary>

| Model | Size | mAP<sup>val<br/>0.5:0.95 | Speed<sup>T4<br/>trt fp16 b1 <br/>(fps) | Speed<sup>T4<br/>trt fp16 b32 <br/>(fps) | Params<br/><sup> (M) | FLOPs<br/><sup> (G) |
| :----------------------------------------------------------- | ---- | :------------------------------------ | --------------------------------------- | ---------------------------------------- | -------------------- | ------------------- |
| [**YOLOv6-N**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6n.pt) | 640 | 35.9<sup>300e</sup><br/>36.3<sup>400e | 802 | 1234 | 4.3 | 11.1 |
| [**YOLOv6-T**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6t.pt) | 640 | 40.3<sup>300e</sup><br/>41.1<sup>400e | 449 | 659 | 15.0 | 36.7 |
| [**YOLOv6-S**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6s.pt) | 640 | 43.5<sup>300e</sup><br/>43.8<sup>400e | 358 | 495 | 17.2 | 44.2 |
| [**YOLOv6-M**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6m.pt) | 640 | 49.5 | 179 | 233 | 34.3 | 82.2 |
| [**YOLOv6-L-ReLU**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6l_relu.pt) | 640 | 51.7 | 113 | 149 | 58.5 | 144.0 |
| [**YOLOv6-L**](https://github.com/meituan/YOLOv6/releases/download/0.2.0/yolov6l.pt) | 640 | 52.5 | 98 | 121 | 58.5 | 144.0 |
- Speed is tested with TensorRT 7.2 on T4.
### Quantized model 🚀

| Model | Size | Precision | mAP<sup>val<br/>0.5:0.95 | Speed<sup>T4<br/>trt b1 <br/>(fps) | Speed<sup>T4<br/>trt b32 <br/>(fps) |
| :-------------------- | ---- | --------- | :----------------------- | ---------------------------------- | ----------------------------------- |
| **YOLOv6-N RepOpt** | 640 | INT8 | 34.8 | 1114 | 1828 |
| **YOLOv6-N** | 640 | FP16 | 35.9 | 802 | 1234 |
| **YOLOv6-T RepOpt** | 640 | INT8 | 39.8 | 741 | 1167 |
| **YOLOv6-T** | 640 | FP16 | 40.3 | 449 | 659 |
| **YOLOv6-S RepOpt** | 640 | INT8 | 43.3 | 619 | 924 |
| **YOLOv6-S** | 640 | FP16 | 43.5 | 377 | 541 |

- Speed is tested with TensorRT 8.4 on T4.
- Precision is figured on models for 300 epochs.

</details>

## Mobile Benchmark
| Model | Size | mAP<sup>val<br/>0.5:0.95 | sm8350<br/><sup>(ms) | mt6853<br/><sup>(ms) | sdm660<br/><sup>(ms) |Params<br/><sup> (M) | FLOPs<br/><sup> (G) |
| :----------------------------------------------------------- | ---- | -------------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------------- |
| [**YOLOv6Lite-S**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6lite_s.pt) | 320*320 | 22.4 | 7.99 | 11.99 | 41.86 | 0.55 | 0.56 |
| [**YOLOv6Lite-M**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6lite_m.pt) | 320*320 | 25.1 | 9.08 | 13.27 | 47.95 | 0.79 | 0.67 |
| [**YOLOv6Lite-L**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6lite_l.pt) | 320*320 | 28.0 | 11.37 | 16.20 | 61.40 | 1.09 | 0.87 |
| [**YOLOv6Lite-L**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6lite_l.pt) | 320*192 | 25.0 | 7.02 | 9.66 | 36.13 | 1.09 | 0.52 |
| [**YOLOv6Lite-L**](https://github.com/meituan/YOLOv6/releases/download/0.4.0/yolov6lite_l.pt) | 224*128 | 18.9 | 3.63 | 4.99 | 17.76 | 1.09 | 0.24 |

<details>
<summary>Table Notes</summary>

- From the perspective of model size and input image ratio, we have built a series of models on the mobile terminal to facilitate flexible applications in different scenarios.
- All checkpoints are trained with 400 epochs without distillation.
- Results of the mAP and speed are evaluated on [COCO val2017](https://cocodataset.org/#download) dataset, and the input resolution is the Size in the table.
- Speed is tested on MNN 2.3.0 AArch64 with 2 threads by arm82 acceleration. The inference warm-up is performed 10 times, and the cycle is performed 100 times.
- Qualcomm 888(sm8350), Dimensity 720(mt6853) and Qualcomm 660(sdm660) correspond to chips with different performances at the high, middle and low end respectively, which can be used as a reference for model capabilities under different chips.
- Refer to [Test NCNN Speed](./docs/Test_NCNN_speed.md) tutorial to reproduce the NCNN speed results of YOLOv6Lite.

</details>

## Quick Start
<details>
<details open>
<summary> Install</summary>


```shell
git clone https://github.com/meituan/YOLOv6
cd YOLOv6
git checkout yolov6-segss
pip install -r requirements.txt
```
</details>



<details>
<summary> Reproduce our results on COCO</summary>

Please refer to [Train COCO Dataset](./docs/Train_coco_data.md).

</details>

<details open>
<summary> Finetune on custom data</summary>
<summary> Training </summary>

Single GPU

```shell
# P5 models
python tools/train.py --batch 32 --conf configs/yolov6s_finetune.py --data data/dataset.yaml --fuse_ab --device 0
# P6 models
python tools/train.py --batch 32 --conf configs/yolov6s6_finetune.py --data data/dataset.yaml --img 1280 --device 0
python tools/train.py --batch 8 --conf configs/yolov6s_finetune.py --data data/coco.yaml --device 0
```

Multi GPUs (DDP mode recommended)

```shell
# P5 models
python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 256 --conf configs/yolov6s_finetune.py --data data/dataset.yaml --fuse_ab --device 0,1,2,3,4,5,6,7
# P6 models
python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 128 --conf configs/yolov6s6_finetune.py --data data/dataset.yaml --img 1280 --device 0,1,2,3,4,5,6,7
python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 64 --conf configs/yolov6s_finetune.py --data data/coco.yaml --device 0,1,2,3,4,5,6,7
```
- fuse_ab: add anchor-based auxiliary branch and use Anchor Aided Training Mode (Not supported on P6 models currently)
- fuse_ab: Not supported in current version
- conf: select config file to specify network/optimizer/hyperparameters. We recommend to apply yolov6n/s/m/l_finetune.py when training on your custom dataset.
- data: prepare dataset and specify dataset paths in data.yaml ( [COCO](http://cocodataset.org), [YOLO format coco labels](https://github.com/meituan/YOLOv6/releases/download/0.1.0/coco2017labels.zip) )
- make sure your dataset structure as follows:
Expand All @@ -169,65 +87,31 @@ YOLOv6 supports different input resolution modes. For details, see [How to Set t

</details>

<details>
<summary>Resume training</summary>

If your training process is corrupted, you can resume training by
```
# single GPU training.
python tools/train.py --resume
# multi GPU training.
python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --resume
```
Above command will automatically find the latest checkpoint in YOLOv6 directory, then resume the training process.

Your can also specify a checkpoint path to `--resume` parameter by
```
# remember to replace /path/to/your/checkpoint/path to the checkpoint path which you want to resume training.
--resume /path/to/your/checkpoint/path
```
This will resume from the specific checkpoint you provide.

</details>

<details open>
<summary> Evaluation</summary>

Reproduce mAP on COCO val2017 dataset with 640×640 or 1280x1280 resolution
Reproduce mAP on COCO val2017 dataset with 640×640 resolution

```shell
# P5 models
python tools/eval.py --data data/coco.yaml --batch 32 --weights yolov6s.pt --task val --reproduce_640_eval
# P6 models
python tools/eval.py --data data/coco.yaml --batch 32 --weights yolov6s6.pt --task val --reproduce_640_eval --img 1280
python tools/eval.py --data data/coco.yaml --batch 32 --weights yolov6s.pt --task val
```
- verbose: set True to print mAP of each classes.
- do_coco_metric: set True / False to enable / disable pycocotools evaluation method.
- do_pr_metric: set True / False to print or not to print the precision and recall metrics.
- config-file: specify a config file to define all the eval params, for example: [yolov6n_with_eval_params.py](configs/experiment/yolov6n_with_eval_params.py)
</details>


<details>
<summary>Inference</summary>

First, download a pretrained model from the YOLOv6 [release](https://github.com/meituan/YOLOv6/releases/tag/0.4.0) or use your trained model to do inference.
First, download a pretrained model from the YOLOv6 [release](https://github.com/meituan/YOLOv6/releases/tag/0.4.1) or use your trained model to do inference.

Second, run inference with `tools/infer.py`

```shell
# P5 models
python tools/infer.py --weights yolov6s.pt --source img.jpg / imgdir / video.mp4
# P6 models
python tools/infer.py --weights yolov6s6.pt --img 1280 1280 --source img.jpg / imgdir / video.mp4
```
If you want to inference on local camera or web camera, you can run:
```shell
# P5 models
python tools/infer.py --weights yolov6s.pt --webcam --webcam-addr 0
# P6 models
python tools/infer.py --weights yolov6s6.pt --img 1280 1280 --webcam --webcam-addr 0
```
`webcam-addr` can be local camera number id or rtsp address.
</details>
Expand Down Expand Up @@ -285,4 +169,4 @@ python tools/infer.py --weights yolov6s6.pt --img 1280 1280 --webcam --webcam-ad
If you have any questions, welcome to join our WeChat group to discuss and exchange.
<p align="center">
<img src="assets/wechat_qrcode.png" align="middle" width = "1000" />
</p>
</p>
Loading

0 comments on commit 8bb4d26

Please sign in to comment.