autoclip: Automated pipeline for AI-driven highlight extraction from Bilibili and YouTube videos

By providing YouTube and Bilibili links, the system handles video downloading, AI analysis, and automated editing in one go. It automates the process of extracting key scenes from long videos and converting them into short-form content.

Using the Tongyi Qianwen LLM to understand video content, it extracts topic-based timelines and assigns a 'highlight score' to each segment. Based on these scores, it cuts out the most interesting parts and automatically generates titles.

The asynchronous processing architecture, built on FastAPI and Celery, delivers real-time progress updates via WebSocket even during large-scale video processing. It can be run with one click in a Docker environment, reducing the burden of complex dependency setup.

Generated clips can be grouped into collections through manual editing or AI recommendations. While Bilibili account integration and automatic upload features are currently under development, the basic pipeline, including local file upload and multilingual subtitle support, is at the completion stage.