%0 Conference Paper %1 Hong:2015:DDN:2969239.2969406 %A Hong, Seunghoon %A Noh, Hyeonwoo %A Han, Bohyung %B Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 1 %C Cambridge, MA, USA %D 2015 %I MIT Press %K decouple segmentation semisupervise %P 1495--1503 %T Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation %U http://dl.acm.org/citation.cfm?id=2969239.2969406 %X We propose a novel deep neural network architecture for semi-supervised semantic segmentation using heterogeneous annotations. Contrary to existing approaches posing semantic segmentation as a single task of region-based classification, our algorithm decouples classification and segmentation, and learns a separate network for each task. In this architecture, labels associated with an image are identified by classification network, and binary segmentation is subsequently performed for each identified label in segmentation network. The decoupled architecture enables us to learn classification and segmentation networks separately based on the training data with image-level and pixel-wise class labels, respectively. It facilitates to reduce search space for segmentation effectively by exploiting class-specific activation maps obtained from bridging layers. Our algorithm shows outstanding performance compared to other semi-supervised approaches with much less training images with strong annotations in PASCAL VOC dataset.

@inproceedings{Hong:2015:DDN:2969239.2969406, abstract = {We propose a novel deep neural network architecture for semi-supervised semantic segmentation using heterogeneous annotations. Contrary to existing approaches posing semantic segmentation as a single task of region-based classification, our algorithm decouples classification and segmentation, and learns a separate network for each task. In this architecture, labels associated with an image are identified by classification network, and binary segmentation is subsequently performed for each identified label in segmentation network. The decoupled architecture enables us to learn classification and segmentation networks separately based on the training data with image-level and pixel-wise class labels, respectively. It facilitates to reduce search space for segmentation effectively by exploiting class-specific activation maps obtained from bridging layers. Our algorithm shows outstanding performance compared to other semi-supervised approaches with much less training images with strong annotations in PASCAL VOC dataset.}, acmid = {2969406}, added-at = {2018-09-17T23:01:42.000+0200}, address = {Cambridge, MA, USA}, author = {Hong, Seunghoon and Noh, Hyeonwoo and Han, Bohyung}, biburl = {https://www.bibsonomy.org/bibtex/24e2bbbc473418a2ea6e6788cfa9811e7/huntstart}, booktitle = {Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 1}, description = {Decoupled deep neural network for semi-supervised semantic segmentation}, interhash = {fa3368d8a8f9f99ee1bbc02301a0e235}, intrahash = {4e2bbbc473418a2ea6e6788cfa9811e7}, keywords = {decouple segmentation semisupervise}, location = {Montreal, Canada}, numpages = {9}, pages = {1495--1503}, publisher = {MIT Press}, series = {NIPS'15}, timestamp = {2018-09-17T23:04:55.000+0200}, title = {Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation}, url = {http://dl.acm.org/citation.cfm?id=2969239.2969406}, year = 2015 }