Overview
We have developed a specialized application for object detection and tracking, utilizing a single camera module. The YOLOv5 trained model was implemented for accurate object detection, and the Multiple Object Tracking (MOT) algorithm was employed for precise tracking.
Goals
The project involved
- setting up Linux and GStreamer on an Nvidia Jetson device. A comprehensive media pipeline was implemented to handle video processing tasks.
- A deepstream plugin was developed to perform accurate object detection and tracking.
- Thorough testing and validation procedures were conducted to ensure the robustness and reliability of the application.
Description
The application consists of multiple modules, one of which is a GStreamer component. This component plays a crucial role in capturing the video source, decoding it, and seamlessly transferring it to Nvidia’s DeepStream custom plugin. This plugin, optimized for high-performance object detection and tracking, efficiently processes the video stream. Additionally, the plugin was designed to upload object tracking metadata to the cloud for further analysis and storage.
Furthermore, it is worth mentioning that the development of this application was carried out using the Nvidia Jetson board, ensuring compatibility and harnessing the board’s capabilities for optimal performance.