The Python AI Plugin is built for developers who want to integrate custom AI models into their live streaming pipeline. Instead of routing video through external services or building your own ingestion pipeline, you implement a small Python class, register it, and your model runs on every live stream that hits Ant Media Server. You get a processed HLS output (with overlays), timestamped detection data via REST API, and a built-in viewer to debug and compare streams.
The plugin has example implementations for pose detection and object detection with YOLO , face detection with OpenCV, and Olama for VLM based AI Processing. You can use this examples as reference when building your own python modules. We will check out Olama integration in our new upcoming blog post.
Top comments (0)