pipeless

Why we included this project

Pipeless takes the serverless model and applies it to video: you write functions that run on each frame, and the framework handles ingestion, scheduling, inference, multi-stream management, and memory reuse itself. That means a detection model that works in a notebook can become an always-on pipeline without hand-assembling a GStreamer or FFmpeg stack, since those sit underneath and Rust does the heavy lifting while your logic stays in Python. It supports common models like YOLO, which makes it a practical fit for edge cameras, live feeds, and media pipelines shipping object detection. Just confirm your input sources and model formats are covered before committing, because you hand over pipeline control in exchange for the callbacks.