SWARD: Stochastic Window-Attention-Based Relational Distillation for Cross-Architectural Semantic Segmentation

• URL: http://arxiv.org/abs/2606.00999v1
• Date: Sun, 31 May 2026 04:29:31 GMT
• Title: SWARD: Stochastic Window-Attention-Based Relational Distillation for Cross-Architectural Semantic Segmentation
• Abstract summary: Large-scale vision foundation models have driven substantial gains on dense prediction tasks such as semantic segmentation.<n>Modern foundation teachers are predominantly transformer-based that encode global context, whereas efficient students are typically convolutional networks with locally biased receptive fields.<n>Existing distillation methods largely assume architectural homogeneity and rely on direct feature mimicry, which fails to bridge this representational gap.<n>We propose SWARD, a knowledge distillation framework that addresses this gap through two complementary mechanisms.
• Score: 1.1617094867403568
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large-scale vision foundation models have driven substantial gains on dense prediction tasks such as semantic segmentation, but their size makes deployment impractical in resource-constrained settings, motivating knowledge distillation as a means of transferring their capabilities to lightweight student networks. However, modern foundation teachers are predominantly transformer-based that encode global context, whereas efficient students are typically convolutional networks with locally biased receptive fields. Existing distillation methods largely assume architectural homogeneity and rely on direct feature mimicry, which fails to bridge this representational gap and neglects the structured spatial dependencies and discriminative organization required for accurate semantic segmentation. In this paper, we propose SWARD, a knowledge distillation framework that addresses this gap through two complementary mechanisms. First, we introduce a Multi-Scale Windowed Attention Distillation (MWAD) module that aligns teacher-student attention-based relations within stochastically shifted window partitions whose offsets are randomly resampled at every training iteration. This removes window boundary bias, and, combined with the multi-scale design, captures both short- and long-range spatial dependencies. Second, we introduce Prototype Discriminative Regularization (PDR), a loss that helps shape the student's feature distribution by enforcing inter-class separation and intra-class compactness, further sharpening the discriminative structure beyond what feature mimicry alone can produce under the student's reduced capacity. Experiments across different vision applications (i.e., urban scene parsing and medical image segmentation) show that SWARD achieves state-of-the-art performance.

Related papers

• Learn to Rank: Visual Attribution by Learning Importance Ranking [58.69028273772474]
  We propose a new approach to generate visual attribution maps for computer vision models.<n>Our method produces dense, pixel-level attributions in a single forward pass with optional, few-step gradient refinement.<n>Our experiments demonstrate consistent quantitative improvements and sharper, boundary-aligned explanations.
  arXiv  Detail & Related papers  (2026-04-07T12:53:22Z)
• Rethinking Vector Field Learning for Generative Segmentation [50.08025820235397]
  Taming diffusion models for generative segmentation has attracted increasing attention.<n>We revisit diffusion segmentation from the perspective of vector field learning.<n>We propose a principled vector field reshaping strategy that augments the learned vanishing velocity field with a detached distance-aware correction term.<n>This correction introduces both attractive and repulsive interactions, enhancing gradient magnitudes near centroids while preserving the original diffusion training framework.
  arXiv  Detail & Related papers  (2026-03-19T17:58:19Z)
• Discriminative Flow Matching Via Local Generative Predictors [0.0]
  We propose a framework that reformulates classification and object detection as a conditional transport process.<n>Our method attaches multiple independent flow predictors to a shared backbone.<n>This architecture provides the flexibility to update blocks either sequentially to minimise activation memory or in parallel to suit different hardware constraints.
  arXiv  Detail & Related papers  (2026-03-14T12:56:29Z)
• UHKD: A Unified Framework for Heterogeneous Knowledge Distillation via Frequency-Domain Representations [5.382357091398666]
  Unified Heterogeneous Knowledge Distillation (UHKD) is proposed as a framework that leverages intermediate features in the frequency domain for cross-architecture transfer.<n>Experiments on CIFAR-100 and ImageNet-1K demonstrate gains of 5.59% and 0.83% over the latest method.
  arXiv  Detail & Related papers  (2025-10-28T06:41:43Z)
• Patch-Level Kernel Alignment for Dense Self-Supervised Learning [7.5866326278176075]
  We introduce Patch-level Kernel Alignment (PaKA), a non-parametric, kernel-based approach that improves the dense representations of pretrained vision encoders with a post-(pre)training.<n>Our framework improves dense representations by conducting a lightweight post-training stage on top of a pretrained model.<n>With only 14 hours of additional training on a single GPU, our method achieves state-of-the-art performance across a range of dense vision benchmarks.
  arXiv  Detail & Related papers  (2025-09-06T05:42:32Z)
• Contrastive Representation Distillation via Multi-Scale Feature Decoupling [2.0401874889991887]
  This work proposes MSDCRD, a model-agnostic distillation framework that systematically decouples global features into multi-scale local features.<n>Extensive experiments demonstrate that MSDCRD achieves superior performance not only in homogeneous teacher-student settings but also in heterogeneous architectures.
  arXiv  Detail & Related papers  (2025-02-09T10:03:18Z)
• Aligning in a Compact Space: Contrastive Knowledge Distillation between Heterogeneous Architectures [4.119589507611071]
  We propose a Low-Frequency Components-based Contrastive Knowledge Distillation (LFCC) framework that significantly enhances the performance of feature-based distillation. Specifically, we designe a set of multi-scale low-pass filters to extract the low-frequency components of intermediate features from both the teacher and student models. We show that LFCC achieves superior performance on the challenging benchmarks of ImageNet-1K and CIFAR-100.
  arXiv  Detail & Related papers  (2024-05-28T18:44:42Z)
• Weakly Supervised Semantic Segmentation via Alternative Self-Dual Teaching [82.71578668091914]
  This paper establishes a compact learning framework that embeds the classification and mask-refinement components into a unified deep model. We propose a novel alternative self-dual teaching (ASDT) mechanism to encourage high-quality knowledge interaction.
  arXiv  Detail & Related papers  (2021-12-17T11:56:56Z)
• Semi-supervised Domain Adaptive Structure Learning [72.01544419893628]
  Semi-supervised domain adaptation (SSDA) is a challenging problem requiring methods to overcome both 1) overfitting towards poorly annotated data and 2) distribution shift across domains. We introduce an adaptive structure learning method to regularize the cooperation of SSL and DA.
  arXiv  Detail & Related papers  (2021-12-12T06:11:16Z)
• HSVA: Hierarchical Semantic-Visual Adaptation for Zero-Shot Learning [74.76431541169342]
  Zero-shot learning (ZSL) tackles the unseen class recognition problem, transferring semantic knowledge from seen classes to unseen ones. We propose a novel hierarchical semantic-visual adaptation (HSVA) framework to align semantic and visual domains. Experiments on four benchmark datasets demonstrate HSVA achieves superior performance on both conventional and generalized ZSL.
  arXiv  Detail & Related papers  (2021-09-30T14:27:50Z)
• Contrastive Distillation on Intermediate Representations for Language Model Compression [89.31786191358802]
  We propose Contrastive Distillation on Intermediate Representations (CoDIR) as a principled knowledge distillation framework. By learning to distinguish positive sample from a large set of negative samples, CoDIR facilitates the student's exploitation of rich information in teacher's hidden layers. CoDIR can be readily applied to compress large-scale language models in both pre-training and finetuning stages, and achieves superb performance on the GLUE benchmark.
  arXiv  Detail & Related papers  (2020-09-29T17:31:43Z)

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.