Chainer implementation of Grad-CAM [1]. Grad-CAM can localize and highlight important region in the image for predicting the concept without changing the model architecture. You can choose AlexNet, VGGNet or ResNet as the model to generate CAM images.
| Grad-CAM | Guided Backpropagation | Guided Grad-CAM | |
|---|---|---|---|
| Boxer (242) |  |
 |
 |
| Tiger Cat (282) |  |
 |
 |
- Chainer
- Cupy (for GPU support)
- OpenCV
python run.py --input images/dog_cat.png --label 242 --layer conv5_3 --gpu 0
python run.py --input images/dog_cat.png --label 282 --layer conv5_3 --gpu 0
- [1] Ramprasaath R. Selvaraju, Abhishek Das, Ramakrishna Vedantam, Michael Cogswell, Devi Parikh, Dhruv Batra, "Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization", https://arxiv.org/abs/1610.02391