%0 Conference Paper %F Oral %T Pushing the boundaries of boundary detection using deep learning %+ Centre de vision numérique (CVN) %+ Organ Modeling through Extraction, Representation and Understanding of Medical Image Content (GALEN) %A Kokkinos, Iasonas %Z Nvidia %< avec comité de lecture %B 4th International Conference on Learning Representations, ICLR 2016 %C San Juan, Puerto Rico %8 2016-05-02 %D 2016 %Z Engineering Sciences [physics]Conference papers %X In this work we show that adapting Deep Convolutional Neural Network training to the task of boundary detection can result in substantial improvements over the current state-of-the-art in boundary detection. Our contributions consist firstly in combining a careful design of the loss for boundary detection training, a multi-resolution architecture and training with external data to improve the detection accuracy of the current state of the art. When measured on the standard Berkeley Segmentation Dataset, we improve theopti-mal dataset scale F-measure from 0.780 to 0.808 - while human performance is at 0.803. We further improve performance to 0.813 by combining deep learning with grouping, integrating the Normalized Cuts technique within a deep network. We also examine the potential of our boundary detector in conjunction with the task of semantic segmentation and demonstrate clear improvements over state-of-the-art systems. Our detector is fully integrated in the popular Caffe framework and processes a 320x420 image in less than a second. %G English %L hal-02432711 %U https://centralesupelec.hal.science/hal-02432711 %~ INRIA %~ INRIA-SACLAY %~ INRIA_TEST %~ CVN %~ TESTALAIN1 %~ CENTRALESUPELEC %~ INRIA2 %~ UNIV-PARIS-SACLAY %~ INRIA-SACLAY-2015 %~ CENTRALESUPELEC-SACLAY %~ GS-ENGINEERING %~ GS-COMPUTER-SCIENCE