update README

SangbumChoi · Jan 12, 2023 · b495484 · b495484
1 parent 7fa9ba9
commit b495484
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 21 deletions.
diff --git a/README.md b/README.md
@@ -97,8 +97,14 @@ pip install -r requirements.txt
 
 
 <details>
-<summary> Training</summary>
+<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>
 
 Single GPU
 
@@ -117,9 +123,9 @@ python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 256
 # 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
 ```
-- fuse_ab: add anchor-based auxiliary branch and use Anchor Unified Training Mode(Not supported on P6 models)
+- fuse_ab: add anchor-based auxiliary branch and use Anchor Unified Training Mode (Not supported on P6 models currently)
 - 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 [COCO](http://cocodataset.org) dataset, [YOLO format coco labels](https://github.com/meituan/YOLOv6/releases/download/0.1.0/coco2017labels.zip) and specify dataset paths in data.yaml
+- 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:
 ```
 ├── coco
@@ -136,8 +142,7 @@ python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 128
 │   ├── README.txt
 ```
 
-
-Reproduce our results on COCO ⭐️ [Train COCO Dataset](./docs/Train_coco_data.md)
+</details>
 
 <details>
 <summary>Resume training</summary>
@@ -159,13 +164,12 @@ Your can also specify a checkpoint path to `--resume` parameter by
 ```
 This will resume from the specific checkpoint you provide.
 
-</details>
 </details>
 
-<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 or 1280x1280 resolution
 
 ```shell
 # P5 models

diff --git a/README_cn.md b/README_cn.md
@@ -14,11 +14,11 @@
 
 
 ## 更新日志
-- [2023.01.06] 发布大分辨率 P6 模型以及对 P5 模型做了全面的升级. ⭐️ [模型指标](#模型指标)
-- [2022.11.04] 发布 [基础版模型](configs/base/README_cn.md) 简化训练部署流程。
+- [2023.01.06] 发布大分辨率 P6 模型以及对 P5 模型做了全面的升级 ⭐️ [模型指标](#模型指标)
+- [2022.11.04] 发布 [基础版模型](configs/base/README_cn.md) 简化训练部署流程
 - [2022.09.06] 定制化的模型量化加速方法 🚀 [量化教程](./tools/qat/README.md)
 - [2022.09.05] 发布 M/L 模型，并且进一步提高了 N/T/S 模型的性能  
-- [2022.06.23] 发布 N/T/S v1.0 版本模型。
+- [2022.06.23] 发布 N/T/S v1.0 版本模型
 
 ## 模型指标
 | 模型                                                       | 输入尺寸 | mAP<sup>val<br/>0.5:0.95              | 速度<sup>T4<br/>trt fp16 b1 <br/>(fps) | 速度<sup>T4<br/>trt fp16 b32 <br/>(fps) | Params<br/><sup> (M) | FLOPs<br/><sup> (G) |
@@ -87,7 +87,14 @@ pip install -r requirements.txt
 </details>
 
 <details>
-<summary> 训练 </summary>
+<summary> 在 COCO 数据集上复现我们的结果</summary>
+
+请参考教程 [训练 COCO 数据集](./docs/Train_coco_data.md).
+
+</details>
+
+<details open>
+<summary> 在自定义数据集上微调模型 </summary>
 
 单卡
 
@@ -106,7 +113,7 @@ python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 256
 # 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
 ```
-- fuse_ab: 增加anchor-based预测分支并使用联合锚点训练模式(P6模型暂不支持)
+- fuse_ab: 增加anchor-based预测分支并使用联合锚点训练模式 (P6模型暂不支持此功能)
 - conf: 配置文件路径，里面包含网络结构、优化器配置、超参数信息。如果您是在自己的数据集训练，我们推荐您使用yolov6n/s/m/l_finetune.py配置文件；
 - data: 数据集配置文件，以 COCO 数据集为例，您可以在 [COCO](http://cocodataset.org) 下载数据, 在这里下载 [YOLO 格式标签](https://github.com/meituan/YOLOv6/releases/download/0.1.0/coco2017labels.zip)；
 - 确保您的数据集按照下面这种格式来组织；
@@ -123,7 +130,7 @@ python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --batch 128
 │   │   ├── val2017
 ```
 
-在COCO数据集复现我们的结果 ⭐️ [训练 COCO 数据集](./docs/Train_coco_data.md)
+</details>
 
 <details>
 <summary>恢复训练</summary>
@@ -147,11 +154,11 @@ python -m torch.distributed.launch --nproc_per_node 8 tools/train.py --resume
 这将从您提供的模型路径恢复训练。
 
 </details>
-</details>
+
 
 <details>
 <summary> 评估</summary>
-在 COCO val2017 数据集上复现我们的结果（输入分辨率 640x640 或 1280x1280） ⭐️
+在 COCO val2017 数据集上复现我们的结果（输入分辨率 640x640 或 1280x1280）
 
 ```shell
 # P5 models
@@ -194,7 +201,7 @@ python tools/infer.py --weights yolov6s6.pt --img 1280 --source img.jpg / imgdir
 <summary> 教程 </summary>
 
 *  [训练 COCO 数据集](./docs/Train_coco_data.md)
-*  [训练自己的数据集](./docs/Train_custom_data.md)
+*  [训练自定义数据集](./docs/Train_custom_data.md)
 *  [测速](./docs/Test_speed.md)
 *  [ YOLOv6 量化教程](./docs/Tutorial%20of%20Quantization.md)
 </details>

diff --git a/docs/Train_custom_data.md b/docs/Train_custom_data.md
@@ -110,21 +110,21 @@ data_aug = dict(
 Single GPU
 
 ```shell
-python tools/train.py --batch 256 --conf configs/yolov6s_finetune.py --data data/data.yaml --device 0
+python tools/train.py --batch 32 --conf configs/yolov6s_finetune.py --data data/dataset.yaml --fuse_ab --device 0
 ```
 
 Multi GPUs (DDP mode recommended)
 
 ```shell
-python -m torch.distributed.launch --nproc_per_node 4 tools/train.py --batch 256 --conf configs/yolov6s_finetune.py --data data/data.yaml --device 0,1,2,3
+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
 ```
 
 
 
 ## 4. Evaluation
 
 ```shell
-python tools/eval.py --data data/data.yaml  --weights output_dir/name/weights/best_ckpt.pt --device 0
+python tools/eval.py --data data/data.yaml  --weights output_dir/name/weights/best_ckpt.pt --task val --device 0
 ```
 
 
@@ -142,5 +142,5 @@ python tools/infer.py --weights output_dir/name/weights/best_ckpt.pt --source im
 Export as [ONNX](https://github.com/meituan/YOLOv6/tree/main/deploy/ONNX) Format
 
 ```shell
-python deploy/ONNX/export_onnx.py --weights output_dir/name/weights/best_ckpt.pt --device 0
+python deploy/ONNX/export_onnx.py --weights output_dir/name/weights/best_ckpt.pt --simplify --device 0
 ```