mlsanityData to 14 Oct 2026

Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks

arXiv 1910.01279 ↗First posted 3 Oct 2019cs.CVcited by 582

Not on the map. Why?

Abstract (arXiv)

Recently, increasing attention has been drawn to the internal mechanisms of convolutional neural networks, and the reason why the network makes specific decisions. In this paper, we develop a novel post-hoc visual explanation method called Score-CAM based on class activation mapping. Unlike previous class activation mapping based approaches, Score-CAM gets rid of the dependence on gradients by obtaining the weight of each activation map through its forward passing score on target class, the final result is obtained by a linear combination of weights and activation maps. We demonstrate that Score-CAM achieves better visual performance and fairness for interpreting the decision making process. Our approach outperforms previous methods on both recognition and localization tasks, it also passes the sanity check. We also indicate its application as debugging tools. Official code has been released.

Key facts. Cited by 582 papers in the map (data through 14 Oct 2026). First posted on arXiv 3 Oct 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 60; 2022: 101; 2023: 106; 2024: 128; 2025: 114; 2026: 73.

582 papers.

Sort:

By area:

  1. 2021Benchmarking and Survey of Explanation Methods for Black Box Modelscited by 136
  2. 2021Interpretable Deep Learning: Interpretation, Interpretability, Trustworthiness, and Beyondcited by 118
  3. 2021HIVE: Evaluating the Human Interpretability of Visual Explanationscited by 105
  4. 2021The effectiveness of feature attribution methods and its correlation with automatic evaluation scorescited by 88
  5. 2022Diffusion Models for Counterfactual Explanationscited by 79
  6. 2021Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU Modelscited by 78
  7. 2022Wave-ViT: Unifying Wavelet and Transformers for Visual Representation Learningcited by 78
  8. 2022Class Re-Activation Maps for Weakly-Supervised Semantic Segmentationcited by 75
  9. 2022MaskTune: Mitigating Spurious Correlations by Forcing to Explorecited by 66
  10. 2023Class Attention Transfer Based Knowledge Distillationcited by 61
  11. 2024Explainable and Interpretable Multimodal Large Language Models: A Comprehensive Surveycited by 52
  12. 2023Learning Bottleneck Concepts in Image Classificationcited by 49
38 more
  1. 2021Towards Better Explanations of Class Activation Mappingcited by 47
  2. 2023Adversarial Counterfactual Visual Explanationscited by 47
  3. 2021There is More than Meets the Eye: Self-Supervised Multi-Object Detection and Tracking with Sound by Distilling Multimodal Knowledgecited by 44
  4. 2022Metrics for saliency map evaluation of deep learning explanation methodscited by 40
  5. 2021Group-CAM: Group Score-Weighted Visual Explanations for Deep Convolutional Networkscited by 36
  6. 2023Visual Explanations of Image-Text Representations via Multi-Modal Information Bottleneck Attributioncited by 34
  7. 2024Gradient based Feature Attribution in Explainable AI: A Technical Reviewcited by 34
  8. 2022Character-Centric Story Visualization via Visual Planning and Token Alignmentcited by 33
  9. 2022Exploring Visual Interpretability for Contrastive Language-Image Pre-trainingcited by 30
  10. 2023Diffusion Models for Counterfactual Generation and Anomaly Detection in Brain Imagescited by 30
  11. 2021Augmenting Convolutional networks with attention-based aggregationcited by 29
  12. 2022Dual Vision Transformercited by 29
  13. 2022Towards Better Understanding Attribution Methodscited by 28
  14. 2022ViT-CX: Causal Explanation of Vision Transformerscited by 28
  15. 2021Visual Explanations from Spiking Neural Networks using Interspike Intervalscited by 27
  16. 2023Segment Anything is A Good Pseudo-label Generator for Weakly Supervised Semantic Segmentationcited by 27
  17. 2021F-CAM: Full Resolution Class Activation Maps via Guided Parametric Upscalingcited by 26
  18. 2021Revisiting The Evaluation of Class Activation Mapping for Explainability: A Novel Metric and Experimental Analysiscited by 25
  19. 2023Opti-CAM: Optimizing saliency maps for interpretabilitycited by 25
  20. 2025Interpretable and Testable Vision Features via Sparse Autoencoderscited by 25
  21. 2022Discovering Transferable Forensic Features for CNN-generated Images Detectioncited by 24
  22. 2022Self-supervised remote sensing feature learning: Learning Paradigms, Challenges, and Future Workscited by 24
  23. 2025FetalCLIP: A Visual-Language Foundation Model for Fetal Ultrasound Image Analysiscited by 24
  24. 2022Shap-CAM: Visual Explanations for Convolutional Neural Networks based on Shapley Valuecited by 23
  25. 2023Overcoming Generic Knowledge Loss with Selective Parameter Updatecited by 23
  26. 2022What Makes a Good Explanation?: A Harmonized View of Properties of Explanationscited by 22
  27. 2023Cross-Layer Retrospective Retrieving via Layer Attentioncited by 22
  28. 2021Robust Models Are More Interpretable Because Attributions Look Normalcited by 21
  29. 2021Self-Interpretable Model with TransformationEquivariant Interpretationcited by 21
  30. 2021Real Masks and Spoof Faces: On the Masked Face Presentation Attack Detectioncited by 20
  31. 2022Abs-CAM: A Gradient Optimization Interpretable Approach for Explanation of Convolutional Neural Networkscited by 20
  32. 2023Boost Vision Transformer with GPU-Friendly Sparsity and Quantizationcited by 20
  33. 2024Interpreting Object-level Foundation Models via Visual Precision Searchcited by 20
  34. 2022ExAID: A Multimodal Explanation Framework for Computer-Aided Diagnosis of Skin Lesionscited by 19
  35. 2022Evaluating Feature Attribution Methods in the Image Domaincited by 19
  36. 2023Visual Explanations via Iterated Integrated Attributionscited by 19
  37. 2021Learning Vision Transformer with Squeeze and Excitation for Facial Expression Recognitioncited by 18
  38. 2021A Study of the Generalizability of Self-Supervised Representationscited by 18

50 most cited of 582 All 582 (JSON)