mirror of https://github.com/open-mmlab/mmpose
88 lines
2.0 KiB
Markdown
88 lines
2.0 KiB
Markdown
# 使用ONNXRuntime进行RTMPose推理
|
|
|
|
本示例展示了如何在Python中用ONNXRuntime推理RTMPose模型。
|
|
|
|
## 准备
|
|
|
|
### 1. 安装onnxruntime推理引擎.
|
|
|
|
选择以下方式之一来安装onnxruntime。
|
|
|
|
- CPU版本
|
|
|
|
```bash
|
|
wget https://github.com/microsoft/onnxruntime/releases/download/v1.8.1/onnxruntime-linux-x64-1.8.1.tgz
|
|
tar -zxvf onnxruntime-linux-x64-1.8.1.tgz
|
|
export ONNXRUNTIME_DIR=$(pwd)/onnxruntime-linux-x64-1.8.1
|
|
export LD_LIBRARY_PATH=$ONNXRUNTIME_DIR/lib:$LD_LIBRARY_PATH
|
|
```
|
|
|
|
- GPU版本
|
|
|
|
```bash
|
|
pip install onnxruntime-gpu==1.8.1
|
|
wget https://github.com/microsoft/onnxruntime/releases/download/v1.8.1/onnxruntime-linux-x64-gpu-1.8.1.tgz
|
|
tar -zxvf onnxruntime-linux-x64-gpu-1.8.1.tgz
|
|
export ONNXRUNTIME_DIR=$(pwd)/onnxruntime-linux-x64-gpu-1.8.1
|
|
export LD_LIBRARY_PATH=$ONNXRUNTIME_DIR/lib:$LD_LIBRARY_PATH
|
|
```
|
|
|
|
### 2. 将模型转换为onnx文件
|
|
|
|
- 安装`mim`工具
|
|
|
|
```bash
|
|
pip install -U openmim
|
|
```
|
|
|
|
- 下载`mmpose`模型
|
|
|
|
```bash
|
|
# choose one rtmpose model
|
|
mim download mmpose --config rtmpose-m_8xb64-270e_coco-wholebody-256x192 --dest .
|
|
```
|
|
|
|
- 克隆`mmdeploy`仓库
|
|
|
|
```bash
|
|
git clone https://github.com/open-mmlab/mmdeploy.git
|
|
```
|
|
|
|
- 将模型转换为onnx文件
|
|
|
|
```bash
|
|
python mmdeploy/tools/deploy.py \
|
|
mmdeploy/configs/mmpose/pose-detection_simcc_onnxruntime_dynamic.py \
|
|
mmpose/rtmpose-m_8xb64-270e_coco-wholebody-256x192.py \
|
|
mmpose/rtmpose-m_simcc-coco-wholebody_pt-aic-coco_270e-256x192-cd5e845c_20230123.pth \
|
|
mmdeploy/demo/resources/human-pose.jpg \
|
|
--work-dir mmdeploy_model/mmpose/ort \
|
|
--device cuda \
|
|
--dump-info
|
|
```
|
|
|
|
## 运行
|
|
|
|
### 安装依赖
|
|
|
|
```bash
|
|
pip install -r requirements.txt
|
|
```
|
|
|
|
### 用法：
|
|
|
|
```bash
|
|
python main.py \
|
|
{ONNX_FILE} \
|
|
{IMAGE_FILE} \
|
|
--device {DEVICE} \
|
|
--save-path {SAVE_PATH}
|
|
```
|
|
|
|
### 参数解释
|
|
|
|
- `ONNX_FILE`: onnx文件的路径
|
|
- `IMAGE_FILE`: 图像文件的路径
|
|
- `DEVICE`: 运行模型的设备，默认为\`cpu'
|
|
- `SAVE_PATH`: 保存输出图像的路径，默认为 "output.jpg"
|