VideoINSTA: Zero-shot Long Video Understanding via Informative Spatial-Temporal Reasoning with LLMs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Liao, Ruotong, Erler, Max, Wang, Huiyu, Zhai, Guangyao, Zhang, Gengyuan, Ma, Yunpu, Tresp, Volker
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917795137060864
author Liao, Ruotong
Erler, Max
Wang, Huiyu
Zhai, Guangyao
Zhang, Gengyuan
Ma, Yunpu
Tresp, Volker
author_facet Liao, Ruotong
Erler, Max
Wang, Huiyu
Zhai, Guangyao
Zhang, Gengyuan
Ma, Yunpu
Tresp, Volker
contents In the video-language domain, recent works in leveraging zero-shot Large Language Model-based reasoning for video understanding have become competitive challengers to previous end-to-end models. However, long video understanding presents unique challenges due to the complexity of reasoning over extended timespans, even for zero-shot LLM-based approaches. The challenge of information redundancy in long videos prompts the question of what specific information is essential for large language models (LLMs) and how to leverage them for complex spatial-temporal reasoning in long-form video analysis. We propose a framework VideoINSTA, i.e. INformative Spatial-TemporAl Reasoning for zero-shot long-form video understanding. VideoINSTA contributes (1) a zero-shot framework for long video understanding using LLMs; (2) an event-based temporal reasoning and content-based spatial reasoning approach for LLMs to reason over spatial-temporal information in videos; (3) a self-reflective information reasoning scheme balancing temporal factors based on information sufficiency and prediction confidence. Our model significantly improves the state-of-the-art on three long video question-answering benchmarks: EgoSchema, NextQA, and IntentQA, and the open question answering dataset ActivityNetQA. The code is released here: https://github.com/mayhugotong/VideoINSTA.
format Preprint
id arxiv_https___arxiv_org_abs_2409_20365
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle VideoINSTA: Zero-shot Long Video Understanding via Informative Spatial-Temporal Reasoning with LLMs
Liao, Ruotong
Erler, Max
Wang, Huiyu
Zhai, Guangyao
Zhang, Gengyuan
Ma, Yunpu
Tresp, Volker
Computer Vision and Pattern Recognition
In the video-language domain, recent works in leveraging zero-shot Large Language Model-based reasoning for video understanding have become competitive challengers to previous end-to-end models. However, long video understanding presents unique challenges due to the complexity of reasoning over extended timespans, even for zero-shot LLM-based approaches. The challenge of information redundancy in long videos prompts the question of what specific information is essential for large language models (LLMs) and how to leverage them for complex spatial-temporal reasoning in long-form video analysis. We propose a framework VideoINSTA, i.e. INformative Spatial-TemporAl Reasoning for zero-shot long-form video understanding. VideoINSTA contributes (1) a zero-shot framework for long video understanding using LLMs; (2) an event-based temporal reasoning and content-based spatial reasoning approach for LLMs to reason over spatial-temporal information in videos; (3) a self-reflective information reasoning scheme balancing temporal factors based on information sufficiency and prediction confidence. Our model significantly improves the state-of-the-art on three long video question-answering benchmarks: EgoSchema, NextQA, and IntentQA, and the open question answering dataset ActivityNetQA. The code is released here: https://github.com/mayhugotong/VideoINSTA.
title VideoINSTA: Zero-shot Long Video Understanding via Informative Spatial-Temporal Reasoning with LLMs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2409.20365