Caffe codes via training and using the model. Contents DHSnet traincode use-caffemodel vgg rank-saliency $$ L = \frac{1}{N} \sum_i \sum_{j\neq y_i} \left[ \max(0, f(x_i; W)j - f(x_i; W){y_i} + \Delta) \right] + \lambda \sum_k\sum_l W_{k,l}^2 $$