RTMPose_TensorRT推理

1.配置环境

假设rtmpose虚拟环境已经适配yolov5的TensorRT推理

conda activate rtmpose
cd ./RTMPose/export_engine/rtmlib-main/
pip install -r requirements.txt

2.转engine

# 首先从官网下载所需要检测的图像大小的onnx文件 链接为：https://github.com/open-mmlab/mmpose/tree/main/projects/rtmpose
conda activate rtmpose
cd ./RTMPose/export_engine/rtmlib-main/
python export.py

3.TensorRT推理测试

修改RTMPose_TRT.py里面的输入输出及其engine路径

之后运行测试python RTMPose_TRT.py即可得到推理出的结果