kjonguk / sa_assign

SA_Assign

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.

Installation

• 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

Usage

To run our system, we provide a script file.

1. Convert Pytorch's pretrained model to ONNX format.

sh ./run.sh store

2. Perform inference using ONNX model with 10 images.

sh ./run.sh run