This video was originally published at the NVIDIA Developer YouTube channel. It is reprinted here with the permission of NVIDIA.
NVIDIA’s DeepStream SDK delivers a complete streaming analytics toolkit for AI-based multi-sensor processing, video, audio and image understanding. This video covers the fundamentals of NVIDIA’s new tracker unified architecture. From the video, you will:
- Learn how to select from the 3 object tracker alternatives (NvDCF, DeepSORT or IOU) or bring your own tracker to DeepStream for vision AI app development.
- Get to know the state machine behind the tracker and what parameters can be configured to optimize the tracker for your specific application.
- Understand how NVIDIA’s state-of-the-art tracker (NvDCF) can compare different configuration parameters, along with the different results and tradeoffs that come from different optimization strategies.
More information about DeepStream tracker:
- Review the DeepStream Low-Level Tracker Comparisons and Tradeoffs for help choosing the best tracker for your application: https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvtracker.html#low-level-tracker-comparisons-and-tradeoffs
- Read the NvMultiObjectTracker Parameter Tuning Guide to learn how to troubleshoot and fine-tune tracker configurations: https://nvda.ws/3NZmbTG
- Check out the Gst-nvtracker plugin documentation for detailed information on how to work with the low-level tracking libraries and how to implement your own: https://nvda.ws/3xyZnTI
- DeepStream get Started resources: https://nvda.ws/39vuk3u
