LongVideo-R1: Smart Navigation for Low-cost Long Video Understanding

TL;DR

This paper introduces LongVideo-R1, a multimodal large language model agent that efficiently navigates long videos by reasoning with high-level cues, reducing redundant processing and improving question-answering accuracy under low computational budgets.

Contribution

The paper presents a novel reasoning-equipped multimodal LLM agent for efficient long video understanding, leveraging hierarchical captions and a two-stage training paradigm including reinforcement learning.

Findings

LongVideo-R1 achieves a superior tradeoff between QA accuracy and efficiency.

The model effectively infers informative video clips using high-level visual cues.

Experiments validate the model's ability to reduce redundant video processing.

Abstract

This paper addresses the critical and underexplored challenge of long video understanding with low computational budgets. We propose LongVideo-R1, an active, reasoning-equipped multimodal large language model (MLLM) agent designed for efficient video context navigation, avoiding the redundancy of exhaustive search. At the core of LongVideo-R1 lies a reasoning module that leverages high-level visual cues to infer the most informative video clip for subsequent processing. During inference, the agent initiates traversal from top-level visual summaries and iteratively refines its focus, immediately halting the exploration process upon acquiring sufficient knowledge to answer the query. To facilitate training, we first extract hierarchical video captions from CGBench, a video corpus with grounding annotations, and guide GPT-5 to generate 33K high-quality chain-of-thought-with-tool trajectories. The LongVideo-R1 agent is fine-tuned upon the Qwen-3-8B model through a two-stage paradigm: supervised fine-tuning (SFT) followed by reinforcement learning (RL), where RL employs a specifically designed reward function to maximize selective and efficient clip navigation. Experiments on multiple long video benchmarks validate the effectiveness of name, which enjoys superior tradeoff between QA accuracy and efficiency. All curated data and source code are provided in the supplementary material and will be made publicly available. Code and data are available at: https://github.com/qiujihao19/LongVideo-R1