Convolutional feature masking for joint object and stuff segmentation

摘要: The topic of semantic segmentation has witnessed considerable progress due to the powerful features learned by convolutional neural networks (CNNs) [13]. current leading approaches for exploit shape information extracting CNN from masked image regions. This strategy introduces artificial boundaries on images and may impact quality extracted features. Besides, operations raw domain require compute thousands a single image, which is time-consuming. In this paper, we propose via masking proposal segments (e.g., super-pixels) are treated as masks feature maps. directly out these maps used train classifiers recognition. We further joint method handle objects “stuff” grass, sky, water) in same framework. State-of-the-art results demonstrated benchmarks PASCAL VOC new PASCAL-CONTEXT, with compelling computational speed.