TimeProVe: A Propose-then-Verify Framework for Efficient Temporal Reasoning in Long Videos

Forum topic · 小凯 · 2026-06-23

Summary

TimeProVe is a cost-efficient hybrid framework for Long Video Question Answering (LVQA), designed to locate sparse, query-relevant evidence in hours-long untrimmed videos. Unlike existing approaches that either densely process videos with large vision-language models (VLMs) at prohibitive computational cost or rely on sparse caption-based reasoning that misses temporally localized, motion-centric evidence, TimeProVe follows a propose-then-verify strategy. Lightweight modules first generate action-grounded answer-evidence hypotheses; an expensive VLM is invoked only for targeted verification. The core Action-based Candidate Evidence (ACE) module converts temporally localized actions into query-conditioned candidate answers and supporting evidence windows via lightweight LLM reasoning. The authors also introduce OpenTSUBench (OTB), an open-ended benchmark for evaluating temporally grounded reasoning in real-world activities of daily living (ADL) scenarios. Experiments show TimeProVe outperforms the strongest baseline by 7.3% on OTB while reducing VLM calls by 75% and inference cost by 93%. Paper: arXiv 2506.18498, published June 23, 2025, in computer vision (CV).

Overview

Field: CV Authors: Arkaprava Sinha, Dominick Reilly, Siddharth Krishnan Published: 2025-06-23 arXiv: 2506.18498

Introduction

Long Video Question Answering (LVQA) requires identifying sparse, query-relevant evidence within hours-long untrimmed videos. Existing approaches either process videos densely with large vision-language models (VLMs), incurring prohibitive computational cost, or rely on sparse caption-based reasoning, which often misses temporally localized and motion-centric evidence.

The TimeProVe Framework

TimeProVe is a cost-efficient hybrid framework for temporally grounded reasoning in long videos, built on a propose-then-verify strategy:

Lightweight modules first generate action-grounded answer–evidence hypotheses.
An expensive VLM is subsequently invoked only for targeted verification.

The core of the framework is the Action-based Candidate Evidence (ACE) module, which converts temporally localized actions into query-conditioned candidate answers and supporting evidence windows using lightweight LLM reasoning.

OpenTSUBench (OTB)

The paper also introduces OpenTSUBench (OTB), an open-ended benchmark for evaluating temporally grounded reasoning in real-world activities of daily living (ADL) scenarios.

Results

Experiments demonstrate that TimeProVe:

Outperforms the strongest baseline by 7.3% on OTB
Reduces VLM calls by 75%
Cuts inference cost by 93%