PastNet: Introducing Physical Inductive Biases for Spatio-temporal Video Prediction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Hao, Xu, Fan, Chen, Chong, Hua, Xian-Sheng, Luo, Xiao, Wang, Haixin
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929674404233216
author Wu, Hao
Xu, Fan
Chen, Chong
Hua, Xian-Sheng
Luo, Xiao
Wang, Haixin
author_facet Wu, Hao
Xu, Fan
Chen, Chong
Hua, Xian-Sheng
Luo, Xiao
Wang, Haixin
contents In this paper, we investigate the challenge of spatio-temporal video prediction task, which involves generating future video frames based on historical spatio-temporal observation streams. Existing approaches typically utilize external information such as semantic maps to improve video prediction accuracy, which often neglect the inherent physical knowledge embedded within videos. Worse still, their high computational costs could impede their applications for high-resolution videos. To address these constraints, we introduce a novel framework called \underline{P}hysics-\underline{a}ssisted \underline{S}patio-\underline{t}emporal \underline{Net}work (PastNet) for high-quality video prediction. The core of PastNet lies in incorporating a spectral convolution operator in the Fourier domain, which efficiently introduces inductive biases from the underlying physical laws. Additionally, we employ a memory bank with the estimated intrinsic dimensionality to discretize local features during the processing of complex spatio-temporal signals, thereby reducing computational costs and facilitating efficient high-resolution video prediction. Extensive experiments on various widely-used spatio-temporal video benchmarks demonstrate the effectiveness and efficiency of the proposed PastNet compared with a range of state-of-the-art methods, particularly in high-resolution scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2305_11421
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle PastNet: Introducing Physical Inductive Biases for Spatio-temporal Video Prediction
Wu, Hao
Xu, Fan
Chen, Chong
Hua, Xian-Sheng
Luo, Xiao
Wang, Haixin
Computer Vision and Pattern Recognition
Artificial Intelligence
14J60 (Primary) 14F05, 14J26 (Secondary)
I.2.6; I.5
In this paper, we investigate the challenge of spatio-temporal video prediction task, which involves generating future video frames based on historical spatio-temporal observation streams. Existing approaches typically utilize external information such as semantic maps to improve video prediction accuracy, which often neglect the inherent physical knowledge embedded within videos. Worse still, their high computational costs could impede their applications for high-resolution videos. To address these constraints, we introduce a novel framework called \underline{P}hysics-\underline{a}ssisted \underline{S}patio-\underline{t}emporal \underline{Net}work (PastNet) for high-quality video prediction. The core of PastNet lies in incorporating a spectral convolution operator in the Fourier domain, which efficiently introduces inductive biases from the underlying physical laws. Additionally, we employ a memory bank with the estimated intrinsic dimensionality to discretize local features during the processing of complex spatio-temporal signals, thereby reducing computational costs and facilitating efficient high-resolution video prediction. Extensive experiments on various widely-used spatio-temporal video benchmarks demonstrate the effectiveness and efficiency of the proposed PastNet compared with a range of state-of-the-art methods, particularly in high-resolution scenarios.
title PastNet: Introducing Physical Inductive Biases for Spatio-temporal Video Prediction
topic Computer Vision and Pattern Recognition
Artificial Intelligence
14J60 (Primary) 14F05, 14J26 (Secondary)
I.2.6; I.5
url https://arxiv.org/abs/2305.11421