Skip to content

Commit

Permalink
Release v0.7.0
Browse files Browse the repository at this point in the history
  • Loading branch information
nepeplwu committed Oct 28, 2020
1 parent 69840ce commit 5d0ec15
Show file tree
Hide file tree
Showing 5 changed files with 21 additions and 20 deletions.
12 changes: 12 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -156,6 +156,18 @@ A: 降低Batch size,使用Group Norm策略;请注意训练过程中当`DEFAU

## 更新日志

* 2020.10.28

**`v0.7.0`**
* 全面支持Paddle2.0-rc动态图模式,推出PaddleSeg[动态图体验版](./dygraph/)
* 发布大量动态图模型,支持11个分割模型,4个骨干网络,3个数据集:
* 分割模型:ANN, BiSeNetV2, DANet, DeeplabV3, DeeplabV3+, FCN, FastSCNN, GCNet, OCRNet, PSPNet, UNet
* 骨干网络:ResNet, HRNet, MobileNetV3, Xception
* 数据集:Cityscapes, ADE20K, Pascal VOC

* 提供高精度骨干网络预训练模型以及基于Cityscapes数据集的语义分割[预训练模型](./dygraph/configs/)。Cityscapes精度超过**82%**


* 2020.08.31

**`v0.6.0`**
Expand Down
10 changes: 5 additions & 5 deletions dygraph/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@

## 数据集

- [x] CityScapes
- [x] Cityscapes
- [x] Pascal VOC
- [x] ADE20K
- [ ] Pascal Context
Expand All @@ -36,7 +36,7 @@

* Python >= 3.6+

由于图像分割模型计算开销大,推荐在GPU版本的PaddlePaddle下使用PaddleSeg.
由于图像分割模型计算开销大,推荐在GPU版本的PaddlePaddle下使用PaddleSeg。推荐安装10.0以上的CUDA环境。


安装教程请见[PaddlePaddle官网](https://www.paddlepaddle.org.cn/documentation/docs/zh/2.0-beta/install/index_cn.html)
Expand All @@ -51,7 +51,7 @@ git clone https://github.com/PaddlePaddle/PaddleSeg
通过以下命令安装python包依赖,请确保在该分支上至少执行过一次以下命令:


```
```shell
cd PaddleSeg/dygraph
export PYTHONPATH=`pwd`
# windows下请执行以下命令
Expand All @@ -60,8 +60,8 @@ pip install -r requirements.txt
```

## 训练
```
python3 train.py --config configs/quick_start/bisenet_optic_disc_512x512_1k.yml
```shell
python train.py --config configs/quick_start/bisenet_optic_disc_512x512_1k.yml
```

## 使用教程
Expand Down
11 changes: 0 additions & 11 deletions dygraph/docs/config.md

This file was deleted.

8 changes: 4 additions & 4 deletions dygraph/docs/data_prepare.md
Original file line number Diff line number Diff line change
Expand Up @@ -40,8 +40,8 @@ python tools/voc_augment.py --voc_path data/VOCdevkit --num_workers 8
**注意** 运行前请确保在dygraph目录下执行过下列命令:
```shell
export PYTHONPATH=`pwd`
# windows下请执行相面的命令
# set PYTHONPATH=%cd%
# windows下请执行相面的命令
# set PYTHONPATH=%cd%
```

## 关于ADE20K数据集
Expand Down Expand Up @@ -81,7 +81,7 @@ export PYTHONPATH=`pwd`
2.标注图像的标签从0,1依次取值,不可间隔。若有需要忽略的像素,则按255进行标注。

可按如下方式对自定义数据集进行配置:
```
```yaml
train_dataset:
type: Dataset
dataset_root: custom_dataset
Expand All @@ -97,4 +97,4 @@ train_dataset:
- type: RandomHorizontalFlip
- type: Normalize
mode: train
```
```
Empty file removed dygraph/docs/infer.md
Empty file.

0 comments on commit 5d0ec15

Please sign in to comment.