CoT-Vid: Dynamic Chain-of-Thought Routing with Self Verification for Training-Free Video Reasoning
This addresses a gap in video reasoning research for AI systems, offering a novel approach that rivals larger proprietary models, though it appears incremental in building on existing chain-of-thought technology.
The paper tackles the problem of complex video reasoning by proposing CoT-Vid, a training-free paradigm that uses dynamic chain-of-thought routing and self-verification, achieving performance gains of 9.3% on Egochema and 5.6% on VideoEspresso compared to its base model.
System2 reasoning is developing rapidly these days with the emergence of Deep- Thinking Models and chain-of-thought technology, which has become a centralized discussion point in the AI community. However, there is a relative gap in the research on complex video reasoning at present. In this work, we propose CoT-Vid, a novel training-free paradigm for the video domain with a multistage complex reasoning design. Distinguishing from existing video LLMs, which rely heavily on perceptual abilities, it achieved surprising performance gain with explicit reasoning mechanism. The paradigm consists of three main components: dynamic inference path routing, problem decoupling strategy, and video self-consistency verification. In addition, we propose a new standard for categorization of video questions. CoT- Vid showed outstanding results on a wide range of benchmarks, and outperforms its base model by 9.3% on Egochema and 5.6% on VideoEspresso, rivalling or even surpassing larger and proprietary models, such as GPT-4V, GPT-4o and Gemini-1.5-flash. Our codebase will be publicly available soon.