Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks
This provides a more reliable tool for researchers and practitioners to debug and understand CNN decisions, though it is incremental as it builds on class activation mapping approaches.
The paper tackles the problem of interpreting convolutional neural network decisions by introducing Score-CAM, a post-hoc visual explanation method that replaces gradient dependence with forward passing scores for activation map weighting, resulting in improved visual performance and fairness, outperforming previous methods on recognition and localization tasks.
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.