mirrors/mmdeploy
1
0
Fork 0
mirror of https://github.com/open-mmlab/mmdeploy.git synced 2025-01-14 08:09:43 +08:00
Code Issues Projects Releases Wiki Activity

[Doc]: MMEdit doc (#153)

Browse Source 
* Add doc

* Remove spaces

* sovle comments

* Resolve comments
This commit is contained in:
Yifan Zhou 2021-10-28 14:33:40 +08:00 committed by GitHub
parent 5fdd75fba1
commit 31446de905
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 60 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
docs/codebases/mmedit.md
View File

@ -1 +1,30 @@
## MMEditing Support
# MMEditing Support
[MMEditing](https://github.com/open-mmlab/mmediting) is an open-source image and video editing toolbox based on PyTorch. It is a part of the [OpenMMLab](https://openmmlab.com/) project.
## MMEditing installation tutorial
Please refer to [official installation guide](https://mmediting.readthedocs.io/en/latest/install.html#installation) to install the codebase.
## List of MMEditing models supported by MMDeploy
| Model | Model Config File (Example) | ONNX Runtime | TensorRT | NCNN | PPL |
| :----: | :----------------------------------------------------------: | :----------: | :------: | :---: | :---: |
| SRCNN | configs/restorers/srcnn/srcnn_x4k915_g1_1000k_div2k.py | Y | Y | N | Y |
| ESRGAN | configs/restorers/esrgan/esrgan_x4c64b23g32_g1_400k_div2k.py | Y | Y | N | Y |
## MMEditing deployment task type
| codebase | task |
| :------: | :--------------: |
| mmedit | super-resolution |
## Reminder
None
## FAQs
1. Why the precision of SRCNN running in TensorRT is lower than in PyTorch?
SRCNN uses bicubic to upsample images. TensorRT doesn't support bicubic operation. Therefore, we replace this operation with bilinear, which may lower the precision.

31
docs_zh_CN/codebases/mmedit.md
View File

@ -1 +1,30 @@
## MMEditing 支持
# MMEditing 支持
[MMEditing](https://github.com/open-mmlab/mmediting) 是一个基于 PyTorch 的开源图像视频编辑工具箱。它是 [OpenMMLab](https://openmmlab.com/) 项目的一部分。
## MMEditing 安装教程
请参考[官方安装指导](https://mmediting.readthedocs.io/en/latest/install.html#installation)进代码库的安装。
## MMDeploy 支持的 MMEditing 模型列表
| 模型 | 模型配置文件（示例） | ONNX Runtime | TensorRT | NCNN | PPL |
| :----: | :----------------------------------------------------------: | :----------: | :------: | :---: | :---: |
| SRCNN | configs/restorers/srcnn/srcnn_x4k915_g1_1000k_div2k.py | Y | Y | N | Y |
| ESRGAN | configs/restorers/esrgan/esrgan_x4c64b23g32_g1_400k_div2k.py | Y | Y | N | Y |
## MMEditing 的部署任务类型
| codebase | task |
| :------: | :--------------: |
| mmedit | super-resolution |
## 注意事项
无
## 常见问题解答
1. 为什么 SRCNN 模型在 TensorRT 上运行的精度低于在 PyTorch 上运行的精度?
SRCNN 使用双三次插值（bicubic）来进行图像上采样。 TensorRT 不支持双三次差值操作。我们用双线性插值（bilinear）替换了该操作，这种替换会降低精度。