Saliency Map for CNN

RV_S

2020 3 6

0 个回答

2 次查看（30 天）

0 个投票

Hello,

I have a googleNet-based image classifier for my dataset and would like to visualize a saliency map of an image. Simply put, I'd like to do a sensitivity analysis of the change in output function with respect to the input image.

This requires me to calculate the change with respect to the weights via backpropagation and I was wondering if there was a simple way to do it in MATLAB. The methodology can be found here. I'm trying to find the MATLAB equivalent of Keras-Vis, but if that's not available, a way to calculate the gradient of the output class with respect to the input image works too.

Thanks!