mlsanityData to 14 Oct 2026

Shape Bias and Robustness Evaluation via Cue Decomposition for Image Classification and Segmentation

arXiv 2503.12453 ↗First posted 16 Mar 2025cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Previous works studied how deep neural networks (DNNs) perceive image content in terms of their biases towards different image cues, such as texture and shape. Previous methods to measure shape and texture biases are typically style-transfer-based and limited to DNNs for image classification. In this work, we provide a new evaluation procedure consisting of 1) a cue-decomposition method that comprises two AI-free data pre-processing methods extracting shape and texture cues, respectively, and 2) a novel cue-decomposition shape bias evaluation metric that leverages the cue-decomposition data. For application purposes we introduce a corresponding cue-decomposition robustness metric that allows for the estimation of the robustness of a DNN w.r.t. image corruptions. In our numerical experiments, our findings for biases in image classification DNNs align with those of previous evaluation metrics. However, our cue-decomposition robustness metric shows superior results in terms of estimating the robustness of DNNs. Furthermore, our results for DNNs on the semantic segmentation datasets Cityscapes and ADE20k for the first time shed light into the biases of semantic segmentation DNNs.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 60 of its 82 references that are in the map. First posted on arXiv 16 Mar 2025.

Builds on

60 of 82 references.

References · 60

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  5. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  6. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  7. 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,859
  8. 2017Squeeze-and-Excitation Networkscited by 4,789
  9. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,444
  10. 2021Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervisioncited by 3,864
  11. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,611
  12. 2022A ConvNet for the 2020scited by 3,579
38 more
  1. 2018Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentationcited by 3,223
  2. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  3. 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,940
  4. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,877
  5. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,546
  6. 2016Pyramid Scene Parsing Networkcited by 2,479
  7. 2021BEiT: BERT Pre-Training of Image Transformerscited by 2,375
  8. 2021Masked-attention Mask Transformer for Universal Image Segmentationcited by 2,351
  9. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
  10. 2020Shortcut Learning in Deep Neural Networkscited by 1,969
  11. 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,850
  12. 2016Adversarial examples in the physical worldcited by 1,769
  13. 2016Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learningcited by 1,512
  14. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
  15. 2020Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformerscited by 1,369
  16. 2019Learning Robust Global Representations by Penalizing Local Predictive Powercited by 1,013
  17. 2020Designing Network Design Spacescited by 960
  18. 2017Network Dissection: Quantifying Interpretability of Deep Visual Representationscited by 708
  19. 2021FLAVA: A Foundational Language And Vision Alignment Modelcited by 649
  20. 2019Segmentation Transformer: Object-Contextual Representations for Semantic Segmentationcited by 576
  21. 2019Using Self-Supervised Learning Can Improve Model Robustness and Uncertaintycited by 514
  22. 2021Intriguing Properties of Vision Transformerscited by 446
  23. 2023EVA-02: A Visual Representation for Neon Genesiscited by 381
  24. 2020Visual Transformers: Token-based Image Representation and Processing for Computer Visioncited by 334
  25. 2021Partial success in closing the gap between human and machine visioncited by 185
  26. 2020Robust and Generalizable Visual Representation Learning via Random Convolutionscited by 142
  27. 2020Disentangled Non-Local Neural Networkscited by 132
  28. 2022MetaFormer Baselines for Visioncited by 132
  29. 2017Dual Path Networkscited by 119
  30. 2021Are Convolutional Neural Networks or Transformers more like human vision?cited by 109
  31. 2021Delving Deep into the Generalization of Vision Transformers under Distribution Shiftscited by 85
  32. 2020Shape-Texture Debiased Neural Network Trainingcited by 76
  33. 2019Interpreting Adversarially Trained Convolutional Neural Networkscited by 75
  34. 2021Shape or Texture: Understanding Discriminative Features in CNNscited by 52
  35. 2023Intriguing properties of generative classifierscited by 52
  36. 2021Robust Contrastive Learning Using Negative Samples with Diminished Semanticscited by 43
  37. 2021Does enhanced shape bias improve neural network robustness to common corruptions?cited by 32
  38. 2024Can We Talk Models Into Seeing the World Differently?cited by 23

50 most cited of 60 All 60 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2025Transferring Styles for Reduced Texture Bias and Improved Robustness in Semantic Segmentation Networkscited by 1
  2. 2026Texture-Shape Bias Balancing for Robust Synthetic-to-Real Semantic Segmentation in Automotive NIR Imagerycited by 0
  3. 2026Investigating the Visual Cues of CNNs for Vascular Segmentation: A Case Study in Microscopy and Fundus Imagingcited by 0