mlsanityData to 14 Oct 2026

The Devil is in Low-Level Features for Cross-Domain Few-Shot Segmentation

arXiv 2503.21150 ↗First posted 27 Mar 2025cs.CV cs.AIcited by 8

Not on the map. Why?

Abstract (arXiv)

Cross-Domain Few-Shot Segmentation (CDFSS) is proposed to transfer the pixel-level segmentation capabilities learned from large-scale source-domain datasets to downstream target-domain datasets, with only a few annotated images per class. In this paper, we focus on a well-observed but unresolved phenomenon in CDFSS: for target domains, particularly those distant from the source domain, segmentation performance peaks at the very early epochs, and declines sharply as the source-domain training proceeds. We delve into this phenomenon for an interpretation: low-level features are vulnerable to domain shifts, leading to sharper loss landscapes during the source-domain training, which is the devil of CDFSS. Based on this phenomenon and interpretation, we further propose a method that includes two plug-and-play modules: one to flatten the loss landscapes for low-level features during source-domain training as a novel sharpness-aware minimization method, and the other to directly supplement target-domain information to the model during target-domain testing by low-level-based calibration. Extensive experiments on four target datasets validate our rationale and demonstrate that our method surpasses the state-of-the-art method in CDFSS signifcantly by 3.71% and 5.34% average MIoU in 1-shot and 5-shot scenarios, respectively.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 37 of its 49 references that are in the map. First posted on arXiv 27 Mar 2025.

Builds on

37 of 49 references.

References · 37

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  4. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  5. 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,859
  6. 2017Prototypical Networks for Few-shot Learningcited by 3,015
  7. 2016Pyramid Scene Parsing Networkcited by 2,479
  8. 2019Similarity of Neural Network Representations Revisitedcited by 1,755
  9. 2020Sharpness-Aware Minimization for Efficiently Improving Generalizationcited by 1,417
  10. 2016Instance Normalization: The Missing Ingredient for Fast Stylizationcited by 1,198
  11. 2019Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)cited by 669
  12. 2019Domain Generalization by Solving Jigsaw Puzzlescited by 450
25 more
  1. 2019PANet: Few-Shot Image Semantic Segmentation with Prototype Alignmentcited by 449
  2. 2018DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Imagescited by 277
  3. 2018Generalizing Across Domains via Cross-Gradient Trainingcited by 268
  4. 2023Personalize Segment Anything Model with One Shotcited by 250
  5. 2021Adaptive Prototype Learning and Allocation for Few-Shot Segmentationcited by 207
  6. 2019FSS-1000: A 1000-Class Dataset for Few-Shot Segmentationcited by 160
  7. 2020Robust and Generalizable Visual Representation Learning via Random Convolutionscited by 142
  8. 2020Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?cited by 135
  9. 2021Few-Shot Segmentation via Cycle-Consistent Transformercited by 133
  10. 2019Learning Robust Representations by Projecting Superficial Statistics Outcited by 123
  11. 2021Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformercited by 116
  12. 2019Large-scale interactive object segmentation with human annotatorscited by 110
  13. 2020Learning to Learn with Variational Information Bottleneck for Domain Generalizationcited by 102
  14. 2022Self-Support Few-Shot Semantic Segmentationcited by 96
  15. 2022Reliability of CKA as a Similarity Measure in Deep Learningcited by 87
  16. 2021Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domaincited by 71
  17. 2022Feature-Proxy Transformer for Few-Shot Segmentationcited by 58
  18. 2024Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Miningcited by 30
  19. 2024Flatten Long-Range Loss Landscapes for Cross-Domain Few-Shot Learningcited by 26
  20. 2024APSeg: Auto-Prompt Network for Cross-Domain Few-Shot Semantic Segmentationcited by 26
  21. 2024Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentationcited by 22
  22. 2024Compositional Few-Shot Class-Incremental Learningcited by 18
  23. 2024Lightweight Frequency Masker for Cross-Domain Few-Shot Semantic Segmentationcited by 17
  24. 2024Domain-Rectifying Adapter for Cross-Domain Few-Shot Segmentationcited by 14
  25. 2020Revisiting Mid-Level Patterns for Cross-Domain Few-Shot Recognitioncited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 6.

8 papers.

  1. 2026Unify the Views: View-Consistent Prototype Learning for Few-Shot Segmentationcited by 1
  2. 2026Training-free Cross-domain Few-shot Segmentation via Robust Semantic Representation and Matchingcited by 1
  3. 2025Inter- and Intra-image Refinement for Few Shot Segmentationcited by 0
  4. 2025Cross-Domain Few-Shot Segmentation via Ordinary Differential Equations over Time Intervalscited by 0
  5. 2026Selective, Regularized, and Calibrated: Harnessing Vision Foundation Models for Cross-Domain Few-Shot Semantic Segmentationcited by 0
  6. 2026Hierarchical Spatial and Channel Aggregation for Cross-domain Few-shot Segmentationcited by 0
  7. 2026Free-Lunch Augmentation by Revisiting Diffusion-Based Data Generation for Cross-Domain Few-Shot Object Detectioncited by 0
  8. 2026The First EgoCross Challenge at EgoVis 2026: Cross-Domain Egocentric Video Question Answeringcited by 0