Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networks

Benchmark Model Rank Results
error-understanding-on-cub-200-2011-resnet-101Grad-CAM++#1Average highest confidence: 0.2647Insertion AUC score: 0.1094
error-understanding-on-cub-200-2011Grad-CAM++#2Average highest confidence (ResNet-101): 0.2647…