DisentangleFormer: Spatial-Channel Decoupling for Multi-Channel Vision

• URL: http://arxiv.org/abs/2512.04314v1
• Date: Wed, 03 Dec 2025 23:03:56 GMT
• Title: DisentangleFormer: Spatial-Channel Decoupling for Multi-Channel Vision
• Authors: Jiashu Liao, Pietro Liò, Marc de Kamps, Duygu Sarikaya,
• Abstract summary: Vision Transformers face a fundamental limitation: standard self-attention jointly processes spatial and channel dimensions.<n>We propose DisentangleFormer, an architecture that achieves robust multi-channel vision representation through principled spatial-channel decoupling.<n>Our design integrates three core components: (1) Parallel Disentanglement: Independently processes spatial-token and channel-token streams, enabling decorrelated feature learning across spatial and spectral dimensions, (2) Squeezed Token Enhancer: An adaptive calibration module that dynamically fuses spatial and channel streams, and (3) Multi-Scale FFN: complementing global attention with multi-scale local context.
• Score: 10.378378296066305
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision Transformers face a fundamental limitation: standard self-attention jointly processes spatial and channel dimensions, leading to entangled representations that prevent independent modeling of structural and semantic dependencies. This problem is especially pronounced in hyperspectral imaging, from satellite hyperspectral remote sensing to infrared pathology imaging, where channels capture distinct biophysical or biochemical cues. We propose DisentangleFormer, an architecture that achieves robust multi-channel vision representation through principled spatial-channel decoupling. Motivated by information-theoretic principles of decorrelated representation learning, our parallel design enables independent modeling of structural and semantic cues while minimizing redundancy between spatial and channel streams. Our design integrates three core components: (1) Parallel Disentanglement: Independently processes spatial-token and channel-token streams, enabling decorrelated feature learning across spatial and spectral dimensions, (2) Squeezed Token Enhancer: An adaptive calibration module that dynamically fuses spatial and channel streams, and (3) Multi-Scale FFN: complementing global attention with multi-scale local context to capture fine-grained structural and semantic dependencies. Extensive experiments on hyperspectral benchmarks demonstrate that DisentangleFormer achieves state-of-the-art performance, consistently outperforming existing models on Indian Pine, Pavia University, and Houston, the large-scale BigEarthNet remote sensing dataset, as well as an infrared pathology dataset. Moreover, it retains competitive accuracy on ImageNet while reducing computational cost by 17.8% in FLOPs. The code will be made publicly available upon acceptance.

Related papers

• Cross-view geo-localization, Image retrieval, Multiscale geometric modeling, Frequency domain enhancement [1.6686955491488273]
  Cross-view geo-localization (CVGL) aims to establish spatial correspondences between images captured from significantly different viewpoints.<n>CVGL remains challenging due to severe geometric asymmetry, texture inconsistency across imaging domains, and the progressive degradation of discriminative local information.<n>This paper proposes the Spatial and Frequency Domain Enhancement Network (SFDE), which leverages complementary representations from spatial and frequency domains.
  arXiv  Detail & Related papers  (2026-03-03T08:25:35Z)
• VFGS-Net: Frequency-Guided State-Space Learning for Topology-Preserving Retinal Vessel Segmentation [14.615144175462051]
  We propose an end-to-end segmentation framework that seamlessly integrates frequency-aware feature enhancement, dual-path convolutional representation learning, and asymmetric spatial state-space modeling within a unified architecture.<n>Our model consistently improves segmentation accuracy for fine vessels, complex branching patterns, and low-contrast regions, highlighting its robustness and clinical potential.
  arXiv  Detail & Related papers  (2026-02-11T16:07:29Z)
• DIFF-MF: A Difference-Driven Channel-Spatial State Space Model for Multi-Modal Image Fusion [51.07069814578009]
  Multi-modal image fusion aims to integrate complementary information from multiple source images to produce high-quality fused images with enriched content.<n>We propose DIFF-MF, a novel difference-driven channel-spatial state space model for multi-modal image fusion.<n>Our method outperforms existing approaches in both visual quality and quantitative evaluation.
  arXiv  Detail & Related papers  (2026-01-09T05:26:54Z)
• Controllable diffusion-based generation for multi-channel biological data [66.44042377817074]
  This work proposes a unified diffusion framework for controllable generation over structured and spatial biological data.<n>We show state-of-the-art performance across both spatial and non-spatial prediction tasks, including protein imputation in IMC and gene-to-protein prediction in single-cell datasets.
  arXiv  Detail & Related papers  (2025-06-24T00:56:21Z)
• Cross Paradigm Representation and Alignment Transformer for Image Deraining [40.66823807648992]
  We propose a novel Cross Paradigm Representation and Alignment Transformer (CPRAformer)<n>Its core idea is the hierarchical representation and alignment, leveraging the strengths of both paradigms to aid image reconstruction.<n>We use two types of self-attention in the Transformer blocks: sparse prompt channel self-attention (SPC-SA) and spatial pixel refinement self-attention (SPR-SA)
  arXiv  Detail & Related papers  (2025-04-23T06:44:46Z)
• Towards Scalable Foundation Model for Multi-modal and Hyperspectral Geospatial Data [14.104497777255137]
  We introduce Low-rank Efficient Spatial-Spectral Vision Transformer with three key innovations.<n>We pretrain LESS ViT using a Hyperspectral Masked Autoencoder framework with integrated positional and channel masking strategies.<n> Experimental results demonstrate that our proposed method achieves competitive performance against state-of-the-art multi-modal geospatial foundation models.
  arXiv  Detail & Related papers  (2025-03-17T05:42:19Z)
• Efficient High-Resolution Visual Representation Learning with State Space Model for Human Pose Estimation [60.80423207808076]
  Capturing long-range dependencies while preserving high-resolution visual representations is crucial for dense prediction tasks such as human pose estimation.<n>We propose the Dynamic Visual State Space (DVSS) block, which augments visual state space models with multi-scale convolutional operations.<n>We build HRVMamba, a novel model for efficient high-resolution representation learning.
  arXiv  Detail & Related papers  (2024-10-04T06:19:29Z)
• Cross-Scope Spatial-Spectral Information Aggregation for Hyperspectral Image Super-Resolution [47.12985199570964]
  We propose a novel cross-scope spatial-spectral Transformer (CST) to investigate long-range spatial and spectral similarities for single hyperspectral image super-resolution. Specifically, we devise cross-attention mechanisms in spatial and spectral dimensions to comprehensively model the long-range spatial-spectral characteristics. Experiments over three hyperspectral datasets demonstrate that the proposed CST is superior to other state-of-the-art methods both quantitatively and visually.
  arXiv  Detail & Related papers  (2023-11-29T03:38:56Z)
• Video Frame Interpolation Transformer [86.20646863821908]
  We propose a Transformer-based video framework that allows content-aware aggregation weights and considers long-range dependencies with the self-attention operations. To avoid the high computational cost of global self-attention, we introduce the concept of local attention into video. In addition, we develop a multi-scale frame scheme to fully realize the potential of Transformers.
  arXiv  Detail & Related papers  (2021-11-27T05:35:10Z)
• Spatial-Temporal Correlation and Topology Learning for Person Re-Identification in Videos [78.45050529204701]
  We propose a novel framework to pursue discriminative and robust representation by modeling cross-scale spatial-temporal correlation. CTL utilizes a CNN backbone and a key-points estimator to extract semantic local features from human body. It explores a context-reinforced topology to construct multi-scale graphs by considering both global contextual information and physical connections of human body.
  arXiv  Detail & Related papers  (2021-04-15T14:32:12Z)
• Channelized Axial Attention for Semantic Segmentation [70.14921019774793]
  We propose the Channelized Axial Attention (CAA) to seamlessly integratechannel attention and axial attention with reduced computationalcomplexity. Our CAA not onlyrequires much less computation resources compared with otherdual attention models such as DANet, but also outperforms the state-of-the-art ResNet-101-based segmentation models on alltested datasets.
  arXiv  Detail & Related papers  (2021-01-19T03:08:03Z)
• Multi-Attention-Network for Semantic Segmentation of Fine Resolution Remote Sensing Images [10.835342317692884]
  The accuracy of semantic segmentation in remote sensing images has been increased significantly by deep convolutional neural networks. This paper proposes a Multi-Attention-Network (MANet) to address these issues. A novel attention mechanism of kernel attention with linear complexity is proposed to alleviate the large computational demand in attention.
  arXiv  Detail & Related papers  (2020-09-03T09:08:02Z)
• Dual Attention GANs for Semantic Image Synthesis [101.36015877815537]
  We propose a novel Dual Attention GAN (DAGAN) to synthesize photo-realistic and semantically-consistent images. We also propose two novel modules, i.e., position-wise Spatial Attention Module (SAM) and scale-wise Channel Attention Module (CAM) DAGAN achieves remarkably better results than state-of-the-art methods, while using fewer model parameters.
  arXiv  Detail & Related papers  (2020-08-29T17:49:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.