# Temporal Localization

VideoMind is a multimodal agent framework that enhances video reasoning capabilities by simulating human thought processes (such as task decomposition, moment localization & verification, and answer synthesis).

CogVLM2-Video is a high-performance video understanding model that achieves state-of-the-art performance in multiple video question-answering tasks, capable of completing video understanding within one minute.