Hi, The weights mentioned in the paper are pixel-wise weight maps. But in code
net_out = self._denseconnect_conv(ae_input, net_name="GuideNet") guide_1 = net_out[:,:,:, 0] * 255 guide_2 = net_out[:,:,:, 1] * 255 guide_3 = net_out[:,:,:, 2] * 255 weight_grid_1 = tf.tile(net_out[:,:,:, 3:4], [1,1,1,3]) weight_grid_2 = tf.tile(net_out[:,:,:, 4:5], [1,1,1,3])
So by reusing the GuideNet to predict weight maps.What I don't understand is how to generate weight maps, what does 3:4,4:5 refer to in the code。I would be grateful if you could answer！
Hi, The weights mentioned in the paper are pixel-wise weight maps. But in code
net_out = self._denseconnect_conv(ae_input, net_name="GuideNet") guide_1 = net_out[:,:,:, 0] * 255 guide_2 = net_out[:,:,:, 1] * 255 guide_3 = net_out[:,:,:, 2] * 255 weight_grid_1 = tf.tile(net_out[:,:,:, 3:4], [1,1,1,3]) weight_grid_2 = tf.tile(net_out[:,:,:, 4:5], [1,1,1,3])So by reusing the GuideNet to predict weight maps.What I don't understand is how to generate weight maps, what does 3:4,4:5 refer to in the code。I would be grateful if you could answer！