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

Haofan Wang, Zifan Wang, Mengnan Du, Fan Yang, Zijian Zhang +3 more
10/3/2019
cs.CV

Official code: haofanwang/Score-CAM (439 stars, MIT, PyTorch, last commit 2022-08-06).

Abstract

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.

View on arXivView PDF

Code Implementations(1)

haofanwang/Score-CAMOfficial100%
PyTorchMIT

Official implementation of Score-CAM in PyTorch

43969PythonApr 13, 20204 years agoMIT
camclass-activation-mapscnn-visualizationcnn-visualization-techniqueexplainability+9 more

Similar Papers

Cluster-CAM: Cluster-Weighted Visual Interpretation of CNNs' Decision in Image Classification

88% match2023

MI CAM: Mutual Information Weighted Activation Mapping for Causal Visual Explanations of Convolutional Neural Networks

88% match2025

Feature Activation Map: Visual Explanation of Deep Learning Models for Image Classification

85% match2023

ViT-ReciproCAM: Gradient and Attention-Free Visual Explanations for Vision Transformer

84% match2023

KPCA-CAM: Visual Explainability of Deep Computer Vision Models using Kernel PCA

84% match1 repo12.9k2024

Abs-CAM: A Gradient Optimization Interpretable Approach for Explanation of Convolutional Neural Networks

84% match2022

Cite this paper

@article{wang2019scorecam,
  title  = {Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks},
  author = {Haofan Wang and Zifan Wang and Mengnan Du and Fan Yang and Zijian Zhang and Sirui Ding and Piotr Mardziel and Xia Hu},
  year   = {2019},
  eprint = {1910.01279},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1910.01279}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1910.01279.svg)](https://codeofpaper.com/paper/1910.01279)

Discussion