This repository contains a tool for running object detection using ONNX model. It provides functions which transforms Pytorch pretrained model to ONNX format and which performs object detection using ONNX model.
- Requirements:
- Python(>= 3.8)
- Pytorch(>= 1.9)
- Install Python libraries:
pip install -r requirements.txt
If you want to use GPU, please install python library:
pip install onnxruntime-gpu
- Other libraries for OpenCV:
apt-get -y install libgl1-mesa-glx libglib2.0-0
To run our system, we provide a script file.
- Convert Pytorch's pretrained model to ONNX format.
sh ./run.sh store
- Perform inference using ONNX model with 10 images.
sh ./run.sh run