We have posted our results on PASCAL VOC Semantic Segmentation Results (VOC2012).
As a (unique research team count), our team (model name: UNIST_GDN) performs 71.8%* (Average Precision %) and is in the 8th position after Oxford, Adelaid, The Chinese University of Hong Kong, UCLA/Google, Inria, Microsoft Research Asia, POSTECH.
– Vladimir Nekrasov, Janghoon Ju and Jaesik Choi, Global Deconvolutional Networks for Semantic Segmentation, arXiv 1602.03930.
As of May 2016, the average precision is 74.0%.
