GOPAgen: Codec-Aware Agentic Long-Video Understanding with Structured Memory
GOPAgen is a novel agentic framework for long-video understanding that integrates video codec primitives and structural memory to overcome limitations in existing methods. It uses a motion agent trained on Group of Pictures (GOPs) to extract motion, builds detailed structural memory through progressive zoom-in, and employs a GOP-Tree reasoning algorithm for efficient long-context processing.