Salvato in:
Dettagli Bibliografici
Autori principali: Yang, Chiao-An, Hachiuma, Ryo, Liu, Sifei, Radhakrishnan, Subhashree, Yeh, Raymond A., Wang, Yu-Chiang Frank, Chen, Min-Hung
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2512.17012
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914465541259264
author Yang, Chiao-An
Hachiuma, Ryo
Liu, Sifei
Radhakrishnan, Subhashree
Yeh, Raymond A.
Wang, Yu-Chiang Frank
Chen, Min-Hung
author_facet Yang, Chiao-An
Hachiuma, Ryo
Liu, Sifei
Radhakrishnan, Subhashree
Yeh, Raymond A.
Wang, Yu-Chiang Frank
Chen, Min-Hung
contents Despite advances in Multimodal LLMs (MLLMs), their ability to reason over 3D structures and temporal dynamics remains limited, constrained by weak 4D perception and temporal understanding. Existing 3D and 4D Video Question Answering (VQA) benchmarks also emphasize static scenes and lack region-level prompting. We tackle these issues by introducing: (a) 4D-RGPT, a specialized MLLM designed to capture 4D representations from video inputs with enhanced temporal perception; (b) Perceptual 4D Distillation (P4D), a training framework that transfers 4D representations from a frozen expert model into 4D-RGPT for comprehensive 4D perception; and (c) R4D-Bench, a benchmark for depth-aware dynamic scenes with region-level prompting, built via a hybrid automated and human-verified pipeline. Our 4D-RGPT achieves notable improvements on both existing 4D VQA benchmarks and the proposed R4D-Bench benchmark.
format Preprint
id arxiv_https___arxiv_org_abs_2512_17012
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation
Yang, Chiao-An
Hachiuma, Ryo
Liu, Sifei
Radhakrishnan, Subhashree
Yeh, Raymond A.
Wang, Yu-Chiang Frank
Chen, Min-Hung
Computer Vision and Pattern Recognition
Despite advances in Multimodal LLMs (MLLMs), their ability to reason over 3D structures and temporal dynamics remains limited, constrained by weak 4D perception and temporal understanding. Existing 3D and 4D Video Question Answering (VQA) benchmarks also emphasize static scenes and lack region-level prompting. We tackle these issues by introducing: (a) 4D-RGPT, a specialized MLLM designed to capture 4D representations from video inputs with enhanced temporal perception; (b) Perceptual 4D Distillation (P4D), a training framework that transfers 4D representations from a frozen expert model into 4D-RGPT for comprehensive 4D perception; and (c) R4D-Bench, a benchmark for depth-aware dynamic scenes with region-level prompting, built via a hybrid automated and human-verified pipeline. Our 4D-RGPT achieves notable improvements on both existing 4D VQA benchmarks and the proposed R4D-Bench benchmark.
title 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.17012