Objects appear to scale differently in natural images. This fact requires methods dealing with object-centric tasks e.g. object proposal to have robust performance over scale variances of objects. In the paper we present a novel segment proposal framework, namely FastMask, which takes advantage of the hierarchical structure in deep convolutional neural network to segment multi-scale objects in one shot. Innovatively, we generalize segment proposal network into three different functional components (body, neck and head). We further propose a weight-shared residual neck module as well as a scale-tolerant attentional head module for multi-scale training and efficient one-shot inference. On MS COCO benchmark, the proposed FastMask outperforms all state-of-the-art segment proposal methods in average recall while keeping 2~5 times faster. More impressively, with a slight trade-off in accuracy, FastMask can segment objects in near real time (~13 fps) at 800$\times$600 resolution images, highlighting its potential in practical applications. Our implementation is available on http://ift.tt/2iKFFAt.
from cs.AI updates on arXiv.org http://ift.tt/2idrHpP
via IFTTT
No comments:
Post a Comment