Test Model Inference Code Repository

This repository provides test code for inference of object detection and object tracking using multiple inference backends with multiple input format sources supported(images, video, stream).

Overview

• All files in the repository, except for those in the tests folder, are shared across all inference backends.

• The tests folder contains subfolders named after specific inference backends. Each subfolder includes the following files:

  • CMakeLists.txt
  • easy-cmake.sh

  The primary differences between inference backends lie in their respective CMakeLists.txt files.

• Currently supported inference backends:

  • yolo-openvino
  • yolo-tensorrt
  • yolo-denglin

---

Usage Instructions

This repository is designed to work alongside other repositories that contain specific inference code (e.g., yolo-openvino). Follow the steps below to set up and run the tests:

Step 1: Clone This Repository

git clone https://github.com/JasonSloan/test-model-infer.git

Step 2: Clone the Target Inference Backend Repository

Build your OpenCV library with video support enabled, and place the compiled files into the 'test-model-infer' directory.

If you are using ubuntu, download the pre-built opencv4.2 library directly from here, and put them into the 'test-model-infer' directory, then set

export LD_LIBRARY_PATH=/path/to/opencv4.2/lib:$LD_LIBRARY_PATH

Step 3: Clone the Target Inference Backend Repository

Clone the repository for the inference backend you wish to test in the same directory as this repository. Ensure that the dependencies for the selected inference backend are correctly installed.

Example:

git clone https://github.com/JasonSloan/yolo-openvino.git

Step 4: Configure the Inference Backend

Modify the CMakeLists.txt file in the subfolder corresponding to the target inference backend (located in the tests folder). Make sure the paths to the inference framework are configured correctly.

Step 5: Run the Test Script

Navigate to the root directory of this repository and execute the easy-cmake.sh script for the specific inference backend.

Example:

cd test-model-infer
sh tests/openvino/easy-cmake.sh

---

This setup allows you to test various inference backends seamlessly with shared code while maintaining backend-specific configurations.