Fusenet.md

Another pretty simple paper.

Let’s say we are dealing with RGBD images. Instead of just adding the Depth as an extra channel, let’s process it with another CNN and add those processed feature maps to the original feature maps.

Done