LNeRV: Disentangled Latent Representation for Semantic-Aware Video Dynamics Modeling
- 29 May 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. PP (10518215) , 1
Abstract
Implicit Neural Representations (INRs) have emerged as a promising paradigm for video coding by enabling continuous signal modeling through coordinate-based neural networks. However, existing video INR methods often model temporal variation either in the pixel domain or with a single dynamic code, which limits their ability to jointly preserve rapidly varying high-frequency details and long-range semantic motion patterns. To address this issue, we propose the Latent-disentangled Neural Representation for Videos (LNeRV), a framework that models video dynamics in latent space using dual dynamic feature grids. Specifically, a denser high-resolution grid is used to capture short-term detail dynamics, while a coarser low-rate grid is used to capture long-range semantic motion patterns. On top of these complementary grids, we introduce a Clustering-based Context Aggregation (CCA) module to combine short-term motion details with long-range temporal context, and a Decoupled Gated Attention (DGA) module as a lightweight gated fusion mechanism to enlarge the receptive field during feature interaction. Extensive experiments demonstrate that LNeRV achieves state-of-the-art video regression performance on the UVG dataset under both PSNR and MS-SSIM metrics. Compared with vanilla NeRV, LNeRV achieves an average PSNR gain of 2.52 dB on UVG, while also showing strong structural fidelity in the compressed rate-distortion setting.Keywords
Funding Information
- National Key Research and Development Program of China (2023YFA1008500)
- National Natural Science Foundation of China (62302128)
- Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (62502116)
- Suzhou Key Core Technology Project (SYG2025118)
This publication has 55 references indexed in Scilit:
- Long-Term Temporal Context Gathering for Neural Video Compression2024
- Latent-INR: A Flexible Framework for Implicit Representations of Videos with Discriminative Semantics2024
- E-NeRV: Expedite Neural Video Representation with Disentangled Spatial-Temporal Context2022
- Multi-modal image synthesis combining content-style adaptive normalization and attentive normalizationComputers & Graphics, 2021
- M-LVC: Multiple Frames Prediction for Learned Video Compression2020
- DVC: An End-To-End Deep Video Compression Framework2019
- Semantic Image Synthesis With Spatially-Adaptive Normalization2019
- Recovering Realistic Texture in Image Super-Resolution by Deep Spatial Feature Transform2018
- A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation2016
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network2016