ParseNet: Looking Wider to See Better

Wei Liu, Andrew Rabinovich, Alexander C. Berg

We present a technique for adding global context to deep convolutional networks for semantic segmentation. The approach is simple, using the average feature for a layer to augment…