Momentor: Advancing Video Large Language Model with Fine-Grained Temporal Reasoning

TL;DR

Momentor is a novel Video-LLM that achieves fine-grained temporal reasoning and localization in videos, overcoming previous models' limitations in segment-level understanding by training on a large-scale, automatically generated dataset.

Contribution

The paper introduces Momentor, a Video-LLM capable of fine-grained temporal reasoning, supported by a new large-scale dataset, Moment-10M, for training and evaluation.

Findings

Momentor outperforms existing models in fine-grained temporal understanding.

It demonstrates strong zero-shot performance on localization tasks.

The dataset enables effective training of segment-level reasoning models.

Abstract

Large Language Models (LLMs) demonstrate remarkable proficiency in comprehending and handling text-based tasks. Many efforts are being made to transfer these attributes to video modality, which are termed Video-LLMs. However, existing Video-LLMs can only capture the coarse-grained semantics and are unable to effectively handle tasks related to comprehension or localization of specific video segments. In light of these challenges, we propose Momentor, a Video-LLM capable of accomplishing fine-grained temporal understanding tasks. To support the training of Momentor, we design an automatic data generation engine to construct Moment-10M, a large-scale video instruction dataset with segment-level instruction data. We train Momentor on Moment-10M, enabling it to perform segment-level reasoning and localization. Zero-shot evaluations on several tasks demonstrate that Momentor excels in fine-grained temporally grounded comprehension and localization.