SRVP: Strong Recollection Video Prediction Model Using Attention-Based Spatiotemporal Correlation Fusion

Yuseon Kim, Kyongseok Park
4/10/2025
cs.CV

No official code found. Best implementation: yuseonk/SRVP (4 stars, last commit 2025-06-22).

Abstract

Video prediction (VP) generates future frames by leveraging spatial representations and temporal context from past frames. Traditional recurrent neural network (RNN)-based models enhance memory cell structures to capture spatiotemporal states over extended durations but suffer from gradual loss of object appearance details. To address this issue, we propose the strong recollection VP (SRVP) model, which integrates standard attention (SA) and reinforced feature attention (RFA) modules. Both modules employ scaled dot-product attention to extract temporal context and spatial correlations, which are then fused to enhance spatiotemporal representations. Experiments on three benchmark datasets demonstrate that SRVP mitigates image quality degradation in RNN-based models while achieving predictive performance comparable to RNN-free architectures.

View on arXivView PDF

Code Implementations(1)

yuseonk/SRVPCommunity100%

This repository contains the official implementation of the paper: "SRVP: Strong Recollection Video Prediction Model Using Attention-Based Spatiotemporal Correlation Fusion" (accepted at CVPR 2025 Precognition Workshop)

40Shell, PythonMar 24, 20251 years ago

Similar Papers

PredRNN: A Recurrent Neural Network for Spatiotemporal Predictive Learning

83% match1 repo5312021

VMRNN: Integrating Vision Mamba and LSTM for Efficient and Accurate Spatiotemporal Forecasting

83% match1 repo1652024

ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction

82% match2026

A neural network that maintains and retrieves memories based on context

82% match2026

Transformer Hawkes Process

82% match1 repo2062020

ARFA: An Asymmetric Receptive Field Autoencoder Model for Spatiotemporal Prediction

82% match2023

Cite this paper

@article{kim2025srvp,
  title  = {SRVP: Strong Recollection Video Prediction Model Using Attention-Based Spatiotemporal Correlation Fusion},
  author = {Yuseon Kim and Kyongseok Park},
  year   = {2025},
  eprint = {2504.08012},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2504.08012v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2504.08012.svg)](https://codeofpaper.com/paper/2504.08012)

Discussion