Where to Look Matters: On-Policy Self-Distillation for Long-Video Understanding

2026-08-26

AI Summary

研究通过引入Clue-OPSD框架，利用线索片段作为特权监督，解决了长视频理解中无关内容干扰的问题，提高了模型预测准确性。

Abstract

Vision-language models (VLMs) have made substantial progress in long-video understanding, with standard backbone models typically answering questions from frames sampled across the full video. However, as videos become longer, the full-video context inevitably contains more question-irrelevant temporal content, which can distract the model from the evidence needed to answer a specific question. We empirically find that focusing the visual input on short annotated clue intervals containing question-relevant evidence consistently improves prediction accuracy across model scales compared with using the corresponding full videos, while requiring fewer input frames. Based on this finding, we introduce Clue-OPSD, a clue-privileged on-policy self-distillation framework for long-video understanding. During training, a full-video student learns from a self-teacher conditioned on the corresponding clue interval by aligning their next-token distributions along student-generated trajectories. Clue-OPSD thus uses clue intervals as privileged supervision without relying on ground-truth answer labels, while requiring no clue annotations or additional modules at inference time. Extensive experiments across multiple long-video understanding benchmarks and Qwen3.5 model scales demonstrate consistent improvements over the corresponding backbone models and strong performance against supervised post-training baselines.

Problem

• long-video understanding
• question-irrelevant temporal content
• prediction accuracy
• visual input
• clue intervals

Innovation

• on-policy self-distillation
• long-video understanding
• clue intervals
• privileged supervision