The output is passed through two convolution layers and then a sigmoid activation is applied to predict the depth of each pixel. |
The output is passed through two convolution layers and then a sigmoid activation is applied to predict the depth of each pixel. |