Keras implementation of GradCAM.
Adapted and optimized code from https://github.com/jacobgil/keras-grad-cam.
Original torch implementation: https://github.com/ramprs/grad-cam.
To run script: python grad_cam.py </path/to/img>
.
Notebook has more details on usage and code to efficiently process multiple images.
Repository contains implementation of three approaches to explain base model predictions:
- Guided Backpropagation;
- Grad-CAM;
- Guided Grad-CAM, which is just multiplication of the first two.
Implementation uses Tensorflow as backend and VGG16 as base model.