PyTorch implementation of various CAM (Class Activation Mapping) models using TorchVision pre-trained CNN Classifiers.
- Dummy-CAM
- Original Implementation (simply average Activation over channels)
- Fake-CAM
- S. Poppi, et al. "Revisiting The Evaluation of Class Activation Mapping for Explainability: A Novel Metric and Experimental Analysis" CVPR 2021.
- CAM (Vanilla-CAM)
- B. Zhou, et al. "Learning Deep Features for Discriminative Localization" CVPR 2016.
- Grad-CAM
- R. Selvaraju, et al. "Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization" ICCV 2017.
- Grad-CAM++
- A. Chattopadhyay, et al. "Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networks" WACV 2018.
- Smooth Grad-CAM++
- D. Omeiza, et al. "Smooth Grad-CAM++: An Enhanced Inference Level Visualization Technique for Deep Convolutional Neural Network Models" IntelliSys 2019.
- XGrad-CAM
- R. Fu, et al. "Axiom-based Grad-CAM: Towards Accurate Visualization and Explanation of CNNs" BMVC 2020.
- IntegratedGrad (InteGrad-CAM)
- M. Sundararajan, et al. "Axiomatic Attribution for Deep Networks" ICML 2017.
- HiRes-CAM
- R. Draelos, et al. "Use HiResCAM instead of Grad-CAM for faithful explanations of convolutional neural networks" arXiv 2020.
- Eigen-CAM
- M. Muhammad, et al. "Eigen-CAM: Class Activation Map using Principal Components" IJCNN 2020.
- Ablation-CAM
- H. Ramaswamy, et al. "Ablation-CAM: Visual Explanations for Deep Convolutional Network via Gradient-free Localization" WACV 2020.
- Score-CAM
- H. Wang, et al. "Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks" CVF 2020.
- Group-CAM
- Q. Zhang, et al. "Group-CAM: Group Score-Weighted Visual Explanations for Deep Convolutional Networks" arXiv 2021.
- Cluster-CAM
- Z. Feng, et al. "Cluster-CAM: Cluster-Weighted Visual Interpretation of CNNs' Decision in Image Classification" arXiv 2023.
- Cosine-CAM
- Original Implementation
- Poly-CAM
- A. Englebert, et al. "Poly-CAM: High resolution class activation map for convolutional neural networks" ICPR 2022.
see usage.ipynb
see cam.ipynb (written in Japanese)
