parent
5869d6cc76
commit
1434efe244
10 changed files with 81 additions and 36 deletions
Binary file not shown.
Binary file not shown.
Before Width: | Height: | Size: 92 KiB After Width: | Height: | Size: 311 KiB |
@ -1,11 +0,0 @@ |
|||||||
PaddleRS develop安装 |
|
||||||
|
|
||||||
github代码会跟随开发进度不断更新,可以安装develop分支的代码使用最新的功能,安装方式如下: |
|
||||||
|
|
||||||
```commandline |
|
||||||
git clone https://github.com/PaddleCV-SIG/PaddleRS |
|
||||||
cd PaddleRS |
|
||||||
git checkout develop |
|
||||||
pip install -r requirements.txt |
|
||||||
python setup.py install |
|
||||||
``` |
|
@ -0,0 +1,42 @@ |
|||||||
|
|
||||||
|
## 环境准备 |
||||||
|
|
||||||
|
- [PaddlePaddle安装](https://www.paddlepaddle.org.cn/install/quick) |
||||||
|
* 版本要求:PaddlePaddle>=2.1.0 |
||||||
|
|
||||||
|
- PaddleRS安装 |
||||||
|
|
||||||
|
|
||||||
|
PaddleRS代码会跟随开发进度不断更新,可以安装develop分支的代码使用最新的功能,安装方式如下: |
||||||
|
|
||||||
|
``` |
||||||
|
git clone https://github.com/PaddleCV-SIG/PaddleRS |
||||||
|
cd PaddleRS |
||||||
|
git checkout develop |
||||||
|
pip install -r requirements.txt |
||||||
|
python setup.py install |
||||||
|
``` |
||||||
|
|
||||||
|
## 开始训练 |
||||||
|
* 在安装PaddleRS后,使用如下命令开始训练,代码会自动下载训练数据, 并均使用单张GPU卡进行训练。 |
||||||
|
|
||||||
|
```commandline |
||||||
|
export CUDA_VISIBLE_DEVICES=0 |
||||||
|
python tutorials/train/semantic_segmentation/deeplabv3p_resnet50_multi_channel.py |
||||||
|
``` |
||||||
|
|
||||||
|
* 若需使用多张GPU卡进行训练,例如使用2张卡时执行: |
||||||
|
|
||||||
|
```commandline |
||||||
|
python -m paddle.distributed.launch --gpus 0,1 tutorials/train/semantic_segmentation/deeplabv3p_resnet50_multi_channel.py |
||||||
|
``` |
||||||
|
使用多卡时,参考[训练参数调整](../../docs/parameters.md)调整学习率和批量大小。 |
||||||
|
|
||||||
|
|
||||||
|
## VisualDL可视化训练指标 |
||||||
|
在模型训练过程,在`train`函数中,将`use_vdl`设为True,则训练过程会自动将训练日志以VisualDL的格式打点在`save_dir`(用户自己指定的路径)下的`vdl_log`目录,用户可以使用如下命令启动VisualDL服务,查看可视化指标 |
||||||
|
```commandline |
||||||
|
visualdl --logdir output/deeplabv3p_resnet50_multi_channel/vdl_log --port 8001 |
||||||
|
``` |
||||||
|
|
||||||
|
服务启动后,使用浏览器打开 https://0.0.0.0:8001 或 https://localhost:8001 |
Loading…
Reference in new issue