mirrors
/
mmdeploy
mirror of https://github.com/open-mmlab/mmdeploy.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[Docs] update 1.x doc links (#1331) 

* update 1.x links

* update docs
pull/1352/head
q.yao 2022-11-10 11:29:16 +08:00 committed by GitHub
parent 6f804979a2
commit 180500d76d
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
25 changed files with 71 additions and 78 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
docs/en/01-how-to-build/android.md
View File

@ -6,7 +6,6 @@
- [Install Dependencies](#install-dependencies)
- [Install Dependencies for SDK](#install-dependencies-for-sdk)
- [Build MMDeploy](#build-mmdeploy)
- [Build Options Spec](#build-options-spec)
- [Build SDK and Demos](#build-sdk-and-demos)
______________________________________________________________________
@ -98,7 +97,7 @@ make -j$(nproc) install
<tr>
<td>OpenJDK </td>
<td>It is necessary for building Java API.</br>
See <a href='https://github.com/open-mmlab/mmdeploy/blob/master/csrc/mmdeploy/apis/java/README.md'> Java API build </a> for building tutorials.
See <a href='https://github.com/open-mmlab/mmdeploy/blob/1.x/csrc/mmdeploy/apis/java/README.md'> Java API build </a> for building tutorials.
</td>
</tr>
</tbody>

2
docs/en/01-how-to-build/build_from_source.md
View File

@ -26,7 +26,7 @@ Note:
- If it fails when `git clone` via `SSH`, you can try the `HTTPS` protocol like this:
```shell
git clone -b master https://github.com/open-mmlab/mmdeploy.git --recursive
git clone -b 1.x https://github.com/open-mmlab/mmdeploy.git --recursive
```
## Build

10
docs/en/01-how-to-build/jetsons.md
View File

@ -163,7 +163,7 @@ conda activate mmdeploy
```shell
sudo apt-get install -y libssl-dev
git clone --branch v1.4.0 https://github.com/open-mmlab/mmcv.git
git clone --branch 2.x https://github.com/open-mmlab/mmcv.git
cd mmcv
MMCV_WITH_OPS=1 pip install -e .
```
@ -223,7 +223,7 @@ It takes about 15 minutes to install ppl.cv on a Jetson Nano. So, please be pati
## Install MMDeploy
```shell
git clone --recursive https://github.com/open-mmlab/mmdeploy.git
git clone -b 1.x --recursive https://github.com/open-mmlab/mmdeploy.git
cd mmdeploy
export MMDEPLOY_DIR=$(pwd)
```
@ -283,15 +283,15 @@ Before running this demo, you need to convert model files to be able to use with
MMDetection is an open source object detection toolbox based on PyTorch
```shell
git clone https://github.com/open-mmlab/mmdetection.git
git clone -b 3.x https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -r requirements/build.txt
pip install -v -e . # or "python setup.py develop"
```
2. Follow [this document](https://github.com/open-mmlab/mmdeploy/blob/master/docs/en/tutorials/how_to_convert_model.md) on how to convert model files.
1. Follow [this document](../02-how-to-run/convert_model.md) on how to convert model files.
For this example, we have used [retinanet_r18_fpn_1x_coco.py](https://github.com/open-mmlab/mmdetection/blob/master/configs/retinanet/retinanet_r18_fpn_1x_coco.py) as the model config, and [this file](https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_r18_fpn_1x_coco/retinanet_r18_fpn_1x_coco_20220407_171055-614fd399.pth) as the corresponding checkpoint file. Also for deploy config, we have used [detection_tensorrt_dynamic-320x320-1344x1344.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py)
For this example, we have used [retinanet_r18_fpn_1x_coco.py](https://github.com/open-mmlab/mmdetection/blob/3.x/configs/retinanet/retinanet_r18_fpn_1x_coco.py) as the model config, and [this file](https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_r18_fpn_1x_coco/retinanet_r18_fpn_1x_coco_20220407_171055-614fd399.pth) as the corresponding checkpoint file. Also for deploy config, we have used [detection_tensorrt_dynamic-320x320-1344x1344.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py)
```shell
python ./tools/deploy.py \

11
docs/en/01-how-to-build/linux-x86_64.md
View File

@ -7,10 +7,8 @@
- [Install Dependencies for SDK](#install-dependencies-for-sdk)
- [Install Inference Engines for MMDeploy](#install-inference-engines-for-mmdeploy)
- [Build MMDeploy](#build-mmdeploy)
- [Build Options Spec](#build-options-spec)
- [Build Model Converter](#build-model-converter)
- [Build Custom Ops](#build-custom-ops)
- [Install Model Converter](#install-model-converter)
- [Install Model Converter](#install-model-converter)
- [Build SDK and Demo](#build-sdk-and-demo)
______________________________________________________________________
@ -71,12 +69,13 @@ conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=11.1 -c pytorch -c c
</td>
</tr>
<tr>
<td>mmcv-full </td>
<td>Install mmcv-full as follows. Refer to the <a href="https://github.com/open-mmlab/mmcv#installation">guide</a> for details.
<td>mmcv </td>
<td>Install mmcv as follows. Refer to the <a href="https://github.com/open-mmlab/mmcv/tree/2.x#installation">guide</a> for details.
<pre><code>
export cu_version=cu111 # cuda 11.1
export torch_version=torch1.8
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/${cu_version}/${torch_version}/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
</code></pre>
</td>
</tr>

5
docs/en/01-how-to-build/macos-arm64.md
View File

@ -34,9 +34,10 @@
Please refer to [get_started](../get_started.md) to install conda.
```bash
# install pytorch & mmcv-full
# install pytorch & mmcv
conda install pytorch==1.9.0 torchvision==0.10.0 -c pytorch
pip install mmcv-full==1.6.0 -f https://download.openmmlab.com/mmcv/dist/cpu/torch1.9.0/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
```
### Install Dependencies for SDK

6
docs/en/01-how-to-build/rockchip.md
View File

@ -42,7 +42,7 @@ It is recommended to create a virtual environment for the project.
conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=11.1 -c pytorch -c conda-forge
```
To work with models from [MMClassification](https://mmclassification.readthedocs.io/en/latest/getting_started.html), you may need to install it additionally.
To work with models from [MMClassification](https://mmclassification.readthedocs.io/en/1.x/get_started.html), you may need to install it additionally.
## Usage
@ -158,7 +158,7 @@ label: 65, score: 0.95
- MMDet models.
YOLOV3 & YOLOX: you may paste the following partition configuration into [detection_rknn_static.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static.py):
YOLOV3 & YOLOX: you may paste the following partition configuration into [detection_rknn_static-320x320.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static-320x320.py):
```python
# yolov3, yolox
@ -173,7 +173,7 @@ label: 65, score: 0.95
])
```
RetinaNet & SSD & FSAF with rknn-toolkit2, you may paste the following partition configuration into [detection_rknn_static.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static.py). Users with rknn-toolkit can directly use default config.
RetinaNet & SSD & FSAF with rknn-toolkit2, you may paste the following partition configuration into [detection_rknn_static-320x320.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static-320x320.py). Users with rknn-toolkit can directly use default config.
```python
# retinanet, ssd

2
docs/en/01-how-to-build/snpe.md
View File

@ -71,7 +71,7 @@ $ python3 -m pip install -e .
## 3. Test the model
Take Resnet-18 as an example. First refer to [documentation to install mmcls](https://github.com/open-mmlab/mmclassification) and use `tools/deploy.py` to convert the model.
Take Resnet-18 as an example. First refer to [documentation to install mmcls](https://github.com/open-mmlab/mmclassification/tree/1.x) and use `tools/deploy.py` to convert the model.
```bash
$ export MODEL_CONFIG=/path/to/mmclassification/configs/resnet/resnet18_8xb16_cifar10.py

11
docs/en/01-how-to-build/windows.md
View File

@ -8,10 +8,8 @@
- [Install Dependencies for SDK](#install-dependencies-for-sdk)
- [Install Inference Engines for MMDeploy](#install-inference-engines-for-mmdeploy)
- [Build MMDeploy](#build-mmdeploy)
- [Build Options Spec](#build-options-spec)
- [Build Model Converter](#build-model-converter)
- [Build Custom Ops](#build-custom-ops)
- [Install Model Converter](#install-model-converter)
- [Install Model Converter](#install-model-converter)
- [Build SDK and Demos](#build-sdk-and-demos)
- [Note](#note)
@ -58,12 +56,13 @@ pip install torch==1.8.0+cu111 torchvision==0.9.0+cu111 torchaudio==0.8.0 -f htt
</td>
</tr>
<tr>
<td>mmcv-full </td>
<td>Install mmcv-full as follows. Refer to the <a href="https://github.com/open-mmlab/mmcv#installation">guide</a> for details.
<td>mmcv </td>
<td>Install mmcv as follows. Refer to the <a href="https://github.com/open-mmlab/mmcv/tree/2.x#installation">guide</a> for details.
<pre><code>
$env:cu_version="cu111"
$env:torch_version="torch1.8.0"
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/${cu_version}/${torch_version}/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
</code></pre>
</td>
</tr>

6
docs/en/02-how-to-run/convert_model.md
View File

@ -11,7 +11,7 @@ Notes:
### Prerequisite
1. Install and build your target backend. You could refer to [ONNXRuntime-install](../05-supported-backends/onnxruntime.md), [TensorRT-install](../05-supported-backends/tensorrt.md), [ncnn-install](../05-supported-backends/ncnn.md), [PPLNN-install](../05-supported-backends/pplnn.md), [OpenVINO-install](../05-supported-backends/openvino.md) for more information.
2. Install and build your target codebase. You could refer to [MMClassification-install](https://github.com/open-mmlab/mmclassification/blob/master/docs/en/install.md), [MMDetection-install](https://github.com/open-mmlab/mmdetection/blob/master/docs/en/get_started.md), [MMSegmentation-install](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/en/get_started.md#installation), [MMOCR-install](https://mmocr.readthedocs.io/en/latest/install.html), [MMEditing-install](https://github.com/open-mmlab/mmediting/blob/master/docs/en/install.md).
2. Install and build your target codebase. You could refer to [MMClassification-install](https://github.com/open-mmlab/mmclassification/blob/1.x/docs/en/get_started.md#installation), [MMDetection-install](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/en/get_started.md), [MMSegmentation-install](https://github.com/open-mmlab/mmsegmentation/blob/1.x/docs/en/get_started.md#installation), [MMOCR-install](https://github.com/open-mmlab/mmocr/blob/1.x/docs/en/get_started/install.md), [MMEditing-install](https://github.com/open-mmlab/mmediting/blob/1.x/docs/en/2_get_started.md#installation).
### Usage
@ -55,8 +55,8 @@ python ./tools/deploy.py \
```bash
python ./tools/deploy.py \
configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py \
$PATH_TO_MMDET/configs/yolo/yolov3_d53_mstrain-608_273e_coco.py \
$PATH_TO_MMDET/checkpoints/yolo/yolov3_d53_mstrain-608_273e_coco.pth \
$PATH_TO_MMDET/configs/yolo/yolov3_d53_8xb8-ms-608-273e_coco.py \
$PATH_TO_MMDET/checkpoints/yolo/yolov3_d53_mstrain-608_273e_coco_20210518_115020-a2c3acb8.pth \
$PATH_TO_MMDET/demo/demo.jpg \
--work-dir work_dir \
--show \

6
docs/en/02-how-to-run/prebuilt_package_windows.md
View File

@ -47,7 +47,7 @@ In order to use the prebuilt package, you need to install some third-party depen
2. Clone the mmdeploy repository
```bash
git clone https://github.com/open-mmlab/mmdeploy.git
git clone -b 1.x https://github.com/open-mmlab/mmdeploy.git
```
:point_right: The main purpose here is to use the configs, so there is no need to compile `mmdeploy`.
@ -55,14 +55,14 @@ In order to use the prebuilt package, you need to install some third-party depen
3. Install mmclassification
```bash
git clone https://github.com/open-mmlab/mmclassification.git
git clone -b 1.x https://github.com/open-mmlab/mmclassification.git
cd mmclassification
pip install -e .
```
4. Prepare a PyTorch model as our example
Download the pth [resnet18_8xb32_in1k_20210831-fbbb1da6.pth](https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth). The corresponding config of the model is [resnet18_8xb32_in1k.py](https://github.com/open-mmlab/mmclassification/blob/master/configs/resnet/resnet18_8xb32_in1k.py)
Download the pth [resnet18_8xb32_in1k_20210831-fbbb1da6.pth](https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth). The corresponding config of the model is [resnet18_8xb32_in1k.py](https://github.com/open-mmlab/mmclassification/blob/1.x/configs/resnet/resnet18_8xb32_in1k.py)
After the above work is done, the structure of the current working directory should be:

2
docs/en/02-how-to-run/write_config.md
View File

@ -177,4 +177,4 @@ detection_tensorrt-int8_dynamic-320x320-1344x1344.py
## 6. How to write model config
According to model's codebase, write the model config file. Model's config file is used to initialize the model, referring to [MMClassification](https://github.com/open-mmlab/mmclassification/blob/master/docs/en/tutorials/config.md), [MMDetection](https://github.com/open-mmlab/mmdetection/blob/master/docs/en/tutorials/config.md), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/en/tutorials/config.md), [MMOCR](https://github.com/open-mmlab/mmocr/blob/main/docs/en/tutorials/config.md), [MMEditing](https://github.com/open-mmlab/mmediting/blob/master/docs/en/config.md).
According to model's codebase, write the model config file. Model's config file is used to initialize the model, referring to [MMClassification](https://github.com/open-mmlab/mmclassification/blob/1.x/docs/en/user_guides/config.md), [MMDetection](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/en/user_guides/config.md), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/blob/1.x/docs/en/user_guides/1_config.md), [MMOCR](https://github.com/open-mmlab/mmocr/blob/1.x/docs/en/user_guides/config.md), [MMEditing](https://github.com/open-mmlab/mmediting/blob/1.x/docs/en/user_guides/1_config.md).

6
docs/en/07-developer-guide/partition_model.md
View File

@ -6,7 +6,7 @@ MMDeploy supports exporting PyTorch models to partitioned onnx models. With this
To support the model partition, we need to add `Mark` nodes in the ONNX model. This could be done with mmdeploy's `@mark` decorator. Note that to make the `mark` work, the marking operation should be included in a rewriting function.
At first, we would mark the model input, which could be done by marking the input tensor `img` in the `forward` method of `BaseDetector` class, which is the parent class of all detector classes. Thus we name this marking point as `detector_forward` and mark the inputs as `input`. Since there could be three outputs for detectors such as `Mask RCNN`, the outputs are marked as `dets`, `labels`, and `masks`. The following code shows the idea of adding mark functions and calling the mark functions in the rewrite. For source code, you could refer to [mmdeploy/codebase/mmdet/models/detectors/base.py](https://github.com/open-mmlab/mmdeploy/blob/86a50e343a3a45d7bc2ba3256100accc4973e71d/mmdeploy/codebase/mmdet/models/detectors/base.py)
At first, we would mark the model input, which could be done by marking the input tensor `img` in the `forward` method of `BaseDetector` class, which is the parent class of all detector classes. Thus we name this marking point as `detector_forward` and mark the inputs as `input`. Since there could be three outputs for detectors such as `Mask RCNN`, the outputs are marked as `dets`, `labels`, and `masks`. The following code shows the idea of adding mark functions and calling the mark functions in the rewrite. For source code, you could refer to [mmdeploy/codebase/mmdet/models/detectors/single_stage.py](https://github.com/open-mmlab/mmdeploy/blob/4fc8828af84281b62be143012cd9f9dafd1e7cc2/mmdeploy/codebase/mmdet/models/detectors/single_stage.py)
```python
from mmdeploy.core import FUNCTION_REWRITER, mark
@ -25,7 +25,7 @@ def base_detector__forward(ctx, self, img, img_metas=None, **kwargs):
return __forward_impl(...)
```
Then, we have to mark the output feature of `YOLOV3Head`, which is the input argument `pred_maps` in `get_bboxes` method of `YOLOV3Head` class. We could add a internal function to only mark the `pred_maps` inside [`yolov3_head__get_bboxes`](https://github.com/open-mmlab/mmdeploy/blob/86a50e343a3a45d7bc2ba3256100accc4973e71d/mmdeploy/codebase/mmdet/models/dense_heads/yolo_head.py#L14) function as following.
Then, we have to mark the output feature of `YOLOV3Head`, which is the input argument `pred_maps` in `get_bboxes` method of `YOLOV3Head` class. We could add a internal function to only mark the `pred_maps` inside [`yolov3_head__get_bboxes`](https://github.com/open-mmlab/mmdeploy/blob/4fc8828af84281b62be143012cd9f9dafd1e7cc2/mmdeploy/codebase/mmdet/models/dense_heads/yolo_head.py#L16) function as following.
```python
from mmdeploy.core import FUNCTION_REWRITER, mark
@ -78,7 +78,7 @@ Once we have marks of nodes and the deployment config with `parition_config` bei
```shell
python tools/torch2onnx.py \
configs/mmdet/detection/yolov3_partition_onnxruntime_static.py \
../mmdetection/configs/yolo/yolov3_d53_mstrain-608_273e_coco.py \
../mmdetection/configs/yolo/yolov3_d53_8xb8-ms-608-273e_coco.py \
https://download.openmmlab.com/mmdetection/v2.0/yolo/yolov3_d53_mstrain-608_273e_coco/yolov3_d53_mstrain-608_273e_coco_20210518_115020-a2c3acb8.pth \
../mmdetection/demo/demo.jpg \
--work-dir ./work-dirs/mmdet/yolov3/ort/partition

3
docs/zh_cn/01-how-to-build/android.md
View File

@ -6,7 +6,6 @@
- [安装依赖包](#安装依赖包)
- [安装 MMDeploy SDK 依赖](#安装-mmdeploy-sdk-依赖)
- [编译 MMDeploy](#编译-mmdeploy)
- [编译选项说明](#编译选项说明)
- [编译 SDK 和 Demos](#编译-sdk-和-demos)
______________________________________________________________________
@ -99,7 +98,7 @@ make -j$(nproc) install
<tr>
<td>OpenJDK </td>
<td>编译Java API之前需要先准备OpenJDK开发环境</br>
请参考 <a href='https://github.com/open-mmlab/mmdeploy/blob/master/csrc/mmdeploy/apis/java/README.md'> Java API 编译 </a> 进行构建.
请参考 <a href='https://github.com/open-mmlab/mmdeploy/blob/1.x/csrc/mmdeploy/apis/java/README.md'> Java API 编译 </a> 进行构建.
</td>
</tr>
</tbody>

2
docs/zh_cn/01-how-to-build/build_from_source.md
View File

@ -27,7 +27,7 @@ git clone -b master git@github.com:open-mmlab/mmdeploy.git --recursive
- 如果以 `SSH` 方式 `git clone` 代码失败，您可以尝试使用 `HTTPS` 协议下载代码：
```bash
git clone -b master https://github.com/open-mmlab/mmdeploy.git MMDeploy
git clone -b 1.x https://github.com/open-mmlab/mmdeploy.git MMDeploy
cd MMDeploy
git submodule update --init --recursive
```

5
docs/zh_cn/01-how-to-build/jetsons.md
View File

@ -149,9 +149,8 @@ conda activate mmdeploy
```shell
sudo apt-get install -y libssl-dev
git clone https://github.com/open-mmlab/mmcv.git
git clone --branch 2.x https://github.com/open-mmlab/mmcv.git
cd mmcv
git checkout v1.4.0
MMCV_WITH_OPS=1 pip install -e .
```
@ -200,7 +199,7 @@ conda activate mmdeploy
## 安装 MMDeploy
```shell
git clone --recursive https://github.com/open-mmlab/mmdeploy.git
git clone -b 1.x --recursive https://github.com/open-mmlab/mmdeploy.git
cd mmdeploy
export MMDEPLOY_DIR=$(pwd)
```

13
docs/zh_cn/01-how-to-build/linux-x86_64.md
View File

@ -8,10 +8,8 @@
- [安装 MMDeploy SDK 依赖](#安装-mmdeploy-sdk-依赖)
- [安装推理引擎](#安装推理引擎)
- [编译 MMDeploy](#编译-mmdeploy)
- [编译选项说明](#编译选项说明)
- [编译安装 Model Converter](#编译安装-model-converter)
- [编译自定义算子](#编译自定义算子)
- [安装 Model Converter](#安装-model-converter)
- [编译 Model Converter](#编译-model-converter)
- [安装 Model Converter](#安装-model-converter)
- [编译 SDK 和 Demos](#编译-sdk-和-demos)
______________________________________________________________________
@ -72,12 +70,13 @@ conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=11.1 -c pytorch -c c
</td>
</tr>
<tr>
<td>mmcv-full </td>
<td>参考如下命令安装 mmcv-full。更多安装方式，可查看 <a href="https://github.com/open-mmlab/mmcv">mmcv 官网</a><br>
<td>mmcv </td>
<td>参考如下命令安装 mmcv。更多安装方式，可查看 <a href="https://github.com/open-mmlab/mmcv">mmcv 官网</a><br>
<pre><code>
export cu_version=cu111 # cuda 11.1
export torch_version=torch1.8
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/${cu_version}/${torch_version}/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
</code></pre>
</td>
</tr>

5
docs/zh_cn/01-how-to-build/macos-arm64.md
View File

@ -37,9 +37,10 @@
参考[get_started](../get_started.md)文档，安装conda。
```bash
# install pytoch & mmcv-full
# install pytoch & mmcv
conda install pytorch==1.9.0 torchvision==0.10.0 -c pytorch
pip install mmcv-full==1.6.0 -f https://download.openmmlab.com/mmcv/dist/cpu/torch1.9.0/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
```
#### 安装 MMDeploy SDK 依赖

6
docs/zh_cn/01-how-to-build/rockchip.md
View File

@ -42,7 +42,7 @@
conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=11.1 -c pytorch -c conda-forge
```
如要使用 [MMClassification](https://mmclassification.readthedocs.io/en/latest/getting_started.html)， 需要用户自己安装使用。
如要使用 [MMClassification](https://mmclassification.readthedocs.io/en/1.x/get_started.html)， 需要用户自己安装使用。
## 使用
@ -158,7 +158,7 @@ label: 65, score: 0.95
- MMDet 模型.
YOLOV3 & YOLOX: 将下面的模型拆分配置写入到 [detection_rknn_static.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static.py):
YOLOV3 & YOLOX: 将下面的模型拆分配置写入到 [detection_rknn_static-320x320.py](https://github.com/open-mmlab/mmdeploy/blob/1.x/configs/mmdet/detection/detection_rknn_static-320x320.py):
```python
# yolov3, yolox
@ -173,7 +173,7 @@ label: 65, score: 0.95
])
```
RetinaNet & SSD & FSAF with rknn-toolkit2, 将下面的模型拆分配置写入到 [detection_rknn_static.py](https://github.com/open-mmlab/mmdeploy/blob/master/configs/mmdet/detection/detection_rknn_static.py)。使用 rknn-toolkit 的用户则不用。
RetinaNet & SSD & FSAF with rknn-toolkit2, 将下面的模型拆分配置写入到 [detection_rknn_static-320x320.py](https://github.com/open-mmlab/mmdeploy/blob/1.x/configs/mmdet/detection/detection_rknn_static-320x320.py)。使用 rknn-toolkit 的用户则不用。
```python
# retinanet, ssd

2
docs/zh_cn/01-how-to-build/snpe.md
View File

@ -75,7 +75,7 @@ tips:
## 三、测试模型
以 Resnet-18 为例。先参照[文档安装 mmcls](https://github.com/open-mmlab/mmclassification)，然后使用 `tools/deploy.py` 转换模型。
以 Resnet-18 为例。先参照[文档安装 mmcls](https://github.com/open-mmlab/mmclassification/tree/1.x)，然后使用 `tools/deploy.py` 转换模型。
```bash
$ export MODEL_CONFIG=/path/to/mmclassification/configs/resnet/resnet18_8xb16_cifar10.py

13
docs/zh_cn/01-how-to-build/windows.md
View File

@ -8,10 +8,8 @@
- [安装 MMDeploy SDK 依赖](#安装-mmdeploy-sdk-依赖)
- [安装推理引擎](#安装推理引擎)
- [编译 MMDeploy](#编译-mmdeploy)
- [编译选项说明](#编译选项说明)
- [编译安装 Model Converter](#编译安装-model-converter)
- [编译自定义算子](#编译自定义算子)
- [安装 Model Converter](#安装-model-converter)
- [编译 Model Converter](#编译-model-converter)
- [安装 Model Converter](#安装-model-converter)
- [编译 SDK 和 Demos](#编译-sdk-和-demos)
- [注意事项](#注意事项)
@ -56,12 +54,13 @@ pip install torch==1.8.0+cu111 torchvision==0.9.0+cu111 torchaudio==0.8.0 -f htt
</td>
</tr>
<tr>
<td>mmcv-full </td>
<td>参考如下命令安装 mmcv-full。更多安装方式，可查看 <a href="https://github.com/open-mmlab/mmcv">mmcv 官网</a><br>
<td>mmcv </td>
<td>参考如下命令安装 mmcv。更多安装方式，可查看 <a href="https://github.com/open-mmlab/mmcv/tree/2.x#installation">mmcv 官网</a><br>
<pre><code>
$env:cu_version="cu111"
$env:torch_version="torch1.8"
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/$env:cu_version/$env:torch_version/index.html
pip install -U openmim
mim install "mmcv>=2.0.0rc1"
</code></pre>
</td>
</tr>

6
docs/zh_cn/02-how-to-run/convert_model.md
View File

@ -26,7 +26,7 @@
### 准备工作
1. 安装您的目标后端。 您可以参考 [ONNXRuntime-install](../05-supported-backends/onnxruntime.md) ，[TensorRT-install](../05-supported-backends/tensorrt.md) ，[ncnn-install](../05-supported-backends/ncnn.md) ，[PPLNN-install](../05-supported-backends/pplnn.md), [OpenVINO-install](../05-supported-backends/openvino.md)。
2. 安装您的目标代码库。 您可以参考 [MMClassification-install](https://github.com/open-mmlab/mmclassification/blob/master/docs/zh_CN/install.md)， [MMDetection-install](https://github.com/open-mmlab/mmdetection/blob/master/docs/zh_cn/get_started.md)， [MMSegmentation-install](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/zh_cn/get_started.md#installation)， [MMOCR-install](https://mmocr.readthedocs.io/zh_CN/latest/install.html)， [MMEditing-install](https://github.com/open-mmlab/mmediting/blob/master/docs/zh_cn/install.md)。
2. 安装您的目标代码库。 您可以参考 [MMClassification-install](https://github.com/open-mmlab/mmclassification/blob/1.x/docs/zh_CN/get_started.md#%E5%AE%89%E8%A3%85)， [MMDetection-install](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/zh_cn/get_started.md)， [MMSegmentation-install](https://github.com/open-mmlab/mmsegmentation/blob/1.x/docs/zh_cn/get_started.md#installation)， [MMOCR-install](https://github.com/open-mmlab/mmocr/blob/1.x/docs/zh_cn/get_started/install.md)， [MMEditing-install](https://github.com/open-mmlab/mmediting/blob/1.x/docs/en/2_get_started.md#installation)。
### 使用方法
@ -70,8 +70,8 @@ python ./tools/deploy.py \
```bash
python ./tools/deploy.py \
configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py \
$PATH_TO_MMDET/configs/yolo/yolov3_d53_mstrain-608_273e_coco.py \
$PATH_TO_MMDET/checkpoints/yolo/yolov3_d53_mstrain-608_273e_coco.pth \
$PATH_TO_MMDET/configs/yolo/yolov3_d53_8xb8-ms-608-273e_coco.py \
$PATH_TO_MMDET/checkpoints/yolo/yolov3_d53_mstrain-608_273e_coco_20210518_115020-a2c3acb8.pth \
$PATH_TO_MMDET/demo/demo.jpg \
--work-dir work_dir \
--show \

6
docs/zh_cn/02-how-to-run/prebuilt_package_windows.md
View File

@ -55,7 +55,7 @@ ______________________________________________________________________
2. 克隆mmdeploy仓库
```bash
git clone https://github.com/open-mmlab/mmdeploy.git
git clone -b 1.x https://github.com/open-mmlab/mmdeploy.git
```
:point_right: 这里主要为了使用configs文件，所以没有加`--recursive`来下载submodule，也不需要编译`mmdeploy`
@ -63,14 +63,14 @@ ______________________________________________________________________
3. 安装mmclassification
```bash
git clone https://github.com/open-mmlab/mmclassification.git
git clone -b 1.x https://github.com/open-mmlab/mmclassification.git
cd mmclassification
pip install -e .
```
4. 准备一个PyTorch的模型文件当作我们的示例
这里选择了[resnet18_8xb32_in1k_20210831-fbbb1da6.pth](https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth)，对应的训练config为[resnet18_8xb32_in1k.py](https://github.com/open-mmlab/mmclassification/blob/master/configs/resnet/resnet18_8xb32_in1k.py)
这里选择了[resnet18_8xb32_in1k_20210831-fbbb1da6.pth](https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth)，对应的训练config为[resnet18_8xb32_in1k.py](https://github.com/open-mmlab/mmclassification/blob/1.x/configs/resnet/resnet18_8xb32_in1k.py)
做好以上工作后，当前工作目录的结构应为：

8
docs/zh_cn/02-how-to-run/write_config.md
View File

@ -6,7 +6,7 @@
<!-- TOC -->
- [如何编写配置文件](#如何编写配置文件)
- [如何写模型转换配置](#如何写模型转换配置)
- [1. 如何编写ONNX配置](#1-如何编写onnx配置)
- [ONNX配置参数说明](#onnx配置参数说明)
- [示例](#示例)
@ -21,8 +21,6 @@
- [5. 部署配置文件命名规则](#5-部署配置文件命名规则)
- [示例](#示例-4)
- [6. 如何编写模型配置文件](#6-如何编写模型配置文件)
- [7. 注意事项](#7-注意事项)
- [8. 常见问题](#8-常见问题)
<!-- TOC -->
@ -106,7 +104,7 @@ codebase_config = dict(type='mmcls', task='Classification')
## 3. 如何编写推理框架配置
推理框架配置主要用于指定模型运行在哪个推理框架，并提供模型在推理框架运行时所需的信息，具体参考 [ONNX Runtime](https://github.com/open-mmlab/mmdeploy/blob/master/docs/en/backends/onnxruntime.md)， [TensorRT](https://github.com/open-mmlab/mmdeploy/blob/master/docs/en/backends/tensorrt.md)， [ncnn](https://github.com/open-mmlab/mmdeploy/blob/master/docs/en/backends/ncnn.md)， [PPLNN](https://github.com/open-mmlab/mmdeploy/blob/master/docs/en/backends/pplnn.md)。
推理框架配置主要用于指定模型运行在哪个推理框架，并提供模型在推理框架运行时所需的信息，具体参考 [ONNX Runtime](../05-supported-backends/onnxruntime.md)， [TensorRT](../05-supported-backends/tensorrt.md)， [ncnn](../05-supported-backends/ncnn.md)， [PPLNN](../05-supported-backends/pplnn.md)。
- `type`: 模型推理框架, 包括 `onnxruntime`， `ncnn`， `pplnn`， `tensorrt`， `openvino`。
@ -189,4 +187,4 @@ detection_tensorrt-int8_dynamic-320x320-1344x1344.py
## 6. 如何编写模型配置文件
请根据模型具体任务的代码库，编写模型配置文件。 模型配置文件用于初始化模型，详情请参考[MMClassification](https://github.com/open-mmlab/mmclassification/blob/master/docs/zh_CN/tutorials/config.md)，[MMDetection](https://github.com/open-mmlab/mmdetection/blob/master/docs/zh_cn/tutorials/config.md)， [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/blob/master/docs/zh_cn/tutorials/config.md)， [MMOCR](https://github.com/open-mmlab/mmocr/blob/main/docs/en/tutorials/config.md)，[MMEditing](https://github.com/open-mmlab/mmediting/blob/master/docs/zh_cn/config.md)。
请根据模型具体任务的代码库，编写模型配置文件。 模型配置文件用于初始化模型，详情请参考[MMClassification](https://github.com/open-mmlab/mmclassification/blob/1.x/docs/zh_CN/user_guides/config.md)，[MMDetection](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/zh_cn/user_guides/config.md)， [MMSegmentation](https://github.com/open-mmlab/mmsegmentation/blob/1.x/docs/zh_cn/user_guides/1_config.md)， [MMOCR](https://github.com/open-mmlab/mmocr/blob/1.x/docs/en/user_guides/config.md)，[MMEditing](https://github.com/open-mmlab/mmediting/blob/1.x/docs/en/user_guides/1_config.md)。

6
docs/zh_cn/07-developer-guide/partition_model.md
View File

@ -6,7 +6,7 @@ MMDeploy 支持将PyTorch模型导出到onnx模型并进行拆分得到多个onn
为了进行图拆分，我们定义了`Mark`类型op，标记模型导出的边界。在实现方法上，采用`mark`装饰器对函数的输入、输出`Tensor`打标记。需要注意的是，我们的标记函数需要在某个重写函数中执行才能生效。
为了对YOLOV3进行拆分，首先我们需要标记模型的输入。这里为了通用性，我们标记检测器父类`BaseDetector`的`forward`方法中的`img` `Tensor`，同时为了支持其他拆分方案，也对`forward`函数的输出进行了标记，分别是`dets`, `labels`和`masks`。下面的代码是截图[mmdeploy/codebase/mmdet/models/detectors/base.py](https://github.com/open-mmlab/mmdeploy/blob/86a50e343a3a45d7bc2ba3256100accc4973e71d/mmdeploy/codebase/mmdet/models/detectors/base.py)中的一部分，可以看出我们使用`mark`装饰器标记了`__forward_impl`函数的输入输出，并在重写函数`base_detector__forward`进行了调用，从而完成了对检测器输入的标记。
为了对YOLOV3进行拆分，首先我们需要标记模型的输入。这里为了通用性，我们标记检测器父类`BaseDetector`的`forward`方法中的`img` `Tensor`，同时为了支持其他拆分方案，也对`forward`函数的输出进行了标记，分别是`dets`, `labels`和`masks`。下面的代码是截图[mmdeploy/codebase/mmdet/models/detectors/single_stage.py](https://github.com/open-mmlab/mmdeploy/blob/4fc8828af84281b62be143012cd9f9dafd1e7cc2/mmdeploy/codebase/mmdet/models/detectors/single_stage.py)中的一部分，可以看出我们使用`mark`装饰器标记了`__forward_impl`函数的输入输出，并在重写函数`base_detector__forward`进行了调用，从而完成了对检测器输入的标记。
```python
from mmdeploy.core import FUNCTION_REWRITER, mark
@ -25,7 +25,7 @@ def base_detector__forward(ctx, self, img, img_metas=None, **kwargs):
return __forward_impl(...)
```
接下来，我们只需要对`YOLOV3Head`中最后一层输出特征`Tensor`进行标记就可以将整个`YOLOV3`模型拆分成两部分。通过查看`mmdet`源码我们可以知道`YOLOV3Head`的`get_bboxes`方法中输入参数`pred_maps`就是我们想要的拆分点，因此可以在重写函数[`yolov3_head__get_bboxes`](https://github.com/open-mmlab/mmdeploy/blob/86a50e343a3a45d7bc2ba3256100accc4973e71d/mmdeploy/codebase/mmdet/models/dense_heads/yolo_head.py#L14)中添加内部函数对`pred_mapes`进行标记，具体参考如下示例代码。值得注意的是，输入参数`pred_maps`是由三个`Tensor`组成的列表，所以我们在onnx模型中添加了三个`Mark`标记节点。
接下来，我们只需要对`YOLOV3Head`中最后一层输出特征`Tensor`进行标记就可以将整个`YOLOV3`模型拆分成两部分。通过查看`mmdet`源码我们可以知道`YOLOV3Head`的`get_bboxes`方法中输入参数`pred_maps`就是我们想要的拆分点，因此可以在重写函数[`yolov3_head__get_bboxes`](https://github.com/open-mmlab/mmdeploy/blob/4fc8828af84281b62be143012cd9f9dafd1e7cc2/mmdeploy/codebase/mmdet/models/dense_heads/yolo_head.py#L16)中添加内部函数对`pred_mapes`进行标记，具体参考如下示例代码。值得注意的是，输入参数`pred_maps`是由三个`Tensor`组成的列表，所以我们在onnx模型中添加了三个`Mark`标记节点。
```python
from mmdeploy.core import FUNCTION_REWRITER, mark
@ -76,7 +76,7 @@ partition_config = dict(
```shell
python tools/torch2onnx.py \
configs/mmdet/detection/yolov3_partition_onnxruntime_static.py \
../mmdetection/configs/yolo/yolov3_d53_mstrain-608_273e_coco.py \
../mmdetection/configs/yolo/yolov3_d53_8xb8-ms-608-273e_coco.py \
https://download.openmmlab.com/mmdetection/v2.0/yolo/yolov3_d53_mstrain-608_273e_coco/yolov3_d53_mstrain-608_273e_coco_20210518_115020-a2c3acb8.pth \
../mmdetection/demo/demo.jpg \
--work-dir ./work-dirs/mmdet/yolov3/ort/partition

4
docs/zh_cn/07-developer-guide/support_new_model.md
View File

@ -24,7 +24,7 @@ def repeat_static(ctx, input, *size):
- `backend`是推理引擎。当模型被导入到引擎的时候，函数会被重载。如果没有给出，重载默认的参数就是重载的参数。如果后端的重载的参数不存在，将会按照预设的默认模式进行重载。
当参数与原始的参数相同时，除了把上下文信息`ctx` 作为第一的参数外，上下文也提供了一些有用的信息，例如:部署的配置`ctx.cfg` 和原始的函数（已经被重载）`ctx.origin_func`。
可参照[这些样例代码](https://github.com/open-mmlab/mmdeploy/blob/master/mmdeploy/codebase/mmcls/models/backbones/shufflenet_v2.py)。
可参照[这些样例代码](https://github.com/open-mmlab/mmdeploy/blob/4d8ea40f55ae9a7ab39d208870fb791de9d46bf9/mmdeploy/codebase/mmcls/models/backbones/shufflenet_v2.py)。
## 模型重载器
@ -87,4 +87,4 @@ def squeeze_default(ctx, g, self, dim=None):
就像函数重载器的`ctx`一样，第一个参数会提供上下文信息。上下文中了一些有用的信息，例如部署配置ctx.cfg和原始功能（已被重载）`ctx.origin_func`。请注意， `ctx.origin_func`只能在`is_pytorch==False`时使用。
[这里](https://github.com/open-mmlab/mmdeploy/tree/master/mmdeploy/pytorch/ops)有很多实现可参考。
[这里](https://github.com/open-mmlab/mmdeploy/tree/6420e2044515ff2052960c0f8bb9e351e6a7f2c2/mmdeploy/pytorch/symbolics)有很多实现可参考。