YOLO C++ API with GUI

This is a YOLO C++ API project which provides a graphical user interface (GUI) for users to customize settings and perform real-time object detection using their webcam. This project has been developed and tested on a Jetson Nano device.

Dependencies

The following dependencies are required to build and run this project. Make sure to install the correct versions as specified below:

• OpenCV (4.x.x)
• Qt (5.15.9)

Setting Up The Models

1. Create a models folder in the root directory.
2. In the models folder, create 3 folders: cfg, names, and weights.
3. Add models you want to use in these 3 folders. For example, if you wanted to use yolov7-tiny, you would need to add yolov7-tiny.weights in the weights folder, yolov7-tiny.cfg in the cfg, and coco.names in the names folder.

How to run the project

1. Clone the repository and navigate to the project directory:

   git clone https://github.com/ethanlaj/yolo-cpp-api.git
   cd yolo-cpp-api
   

2. Compile the project:

   make
   

3. Run the compiled binary:

   ./yolo_detect
   

4. The GUI will open, allowing you to choose a specific model to use. This project only supports .weights formatted models.

5. Click the "Start" button to begin real-time object detection using your webcam. Detected objects will be displayed with bounding boxes and labels.

GUI Features

• Ability to use different models
• While running, can customize object detection settings such as confidence and NMS thresholds.
• Start and stop real-time object detection.
• Display detected objects with bounding boxes and labels.

Troubleshooting

Camera frames not capturing

If frames cannot be captured by the camera, try restarting the nvargus-daemon by running the following command:

sudo systemctl restart nvargus-daemon