ParseNet: Looking Wider to See Better

Wei Liu, Andrew Rabinovich, Alexander C. Berg

CMT Id: 240

Abstract: We present a technique for adding global context to fully convolutional networks for semantic segmentation. The approach is simple, using the average feature for a layer to augment the features at each location. In addition, we study several idiosyncrasies of training, significantly increasing the performance of baseline networks (e.g. from FCN~\cite{long2014fully}). When we add our proposed global feature, and a technique for learning normalization parameters, accuracy increases consistently even over our improved versions of the baselines. Our proposed approach, ParseNet, achieves state-of-the-art performance on SiftFlow and PASCAL-Context with small additional computational cost over baselines, and near state-of-the-art performance on PASCAL VOC 2012 semantic segmentation with a simple approach. Code is available at \url{https://github.com/weiliu89/caffe/tree/fcn} .

Conflicts: cs.unc.edu, magicleap.com