noureddinekhiati/seg-gradientcam

Seg-GradCAM: Towards Interpretable Semantic Segmentation via Gradient-weighted Class Activation Mapping

This repository contains the Keras/Tensorflow implementation of Segmentation Gradient CAM. The preprint of the paper is available at Towards Interpretable Semantic Segmentation via Gradient-weighted Class Activation Mapping. Though numerous interpretable approaches have been proposed to explain the decision of the classification models, the interpretation of image segmentation still remains largely unexplored. To this end authors proposed an approach named SEG-GRAD-CAM (an extension of Grad-CAM) for interpreting semantic segmentation. The Pytorch Implementation for this proposed pparoach is also available.

Implementation

Code: Google Colab Notebook

Results