MDP3: A Training-free Approach for List-wise Frame Selection in Video-LLMs

TL;DR

This paper introduces MDP3, a training-free, model-agnostic method for selecting relevant and diverse video frames based on query relevance, list-wise diversity, and sequentiality, improving video understanding in large language models.

Contribution

The paper proposes MDP3, a novel approach combining Markov decision processes and determinantal point processes for effective, training-free frame selection in Video-LLMs, addressing limitations of existing methods.

Findings

MDP3 achieves a rac{1-1/e}{ ext{approximate solution for NP-hard problem}}.

Empirical results show MDP3 significantly outperforms existing frame selection methods.

The method is efficient and robust across different video datasets.

Abstract

Video large language models (Video-LLMs) have made significant progress in understanding videos. However, processing multiple frames leads to lengthy visual token sequences, presenting challenges such as the limited context length cannot accommodate the entire video, and the inclusion of irrelevant frames hinders visual perception. Hence, effective frame selection is crucial. This paper emphasizes that frame selection should follow three key principles: query relevance, list-wise diversity, and sequentiality. Existing methods, such as uniform frame sampling and query-frame matching, do not capture all of these principles. Thus, we propose Markov decision determinantal point process with dynamic programming (MDP3) for frame selection, a training-free and model-agnostic method that can be seamlessly integrated into existing Video-LLMs. Our method first estimates frame similarities conditioned on the query using a conditional Gaussian kernel within the reproducing kernel Hilbert space~(RKHS). We then apply the determinantal point process~(DPP) to the similarity matrix to capture both query relevance and list-wise diversity. To incorporate sequentiality, we segment the video and apply DPP within each segment, conditioned on the preceding segment selection, modeled as a Markov decision process~(MDP) for allocating selection sizes across segments. Theoretically, MDP3 provides a \((1 - 1/e)\)-approximate solution to the NP-hard list-wise frame selection problem with pseudo-polynomial time complexity, demonstrating its efficiency. Empirically, MDP3 significantly outperforms existing methods, verifying its effectiveness and robustness.