$S^4$Net: Single Stage Salient-Instance Segmentation
Ruochen Fan
and
Qibin Hou
and
Ming-Ming Cheng
and
Tai-Jiang Mu
and
Shi-Min Hu
arXiv e-Print archive - 2017 via Local arXiv
Keywords:
cs.CV
First published: 2017/11/21 (7 years ago) Abstract: In this paper, we consider an interesting vision problem---salient instance
segmentation. Other than producing approximate bounding boxes, our network also
outputs high-quality instance-level segments. Taking into account the
category-independent property of each target, we design a single stage salient
instance segmentation framework, with a novel segmentation branch. Our new
branch regards not only local context inside each detection window but also its
surrounding context, enabling us to distinguish the instances in the same scope
even with obstruction. Our network is end-to-end trainable and runs at a fast
speed (40 fps when processing an image with resolution $320 \times 320$). We
evaluate our approach on a public available benchmark and show that it
outperforms other alternative solutions. In addition, we also provide a
thorough analysis of the design choices to help readers better understand the
functions of each part of our network. To facilitate the development of this
area, our code will be available at \url{https://github.com/RuochenFan/S4Net}.
It's like mask rcnn but for salient instances.
code will be available at https://github.com/RuochenFan/S4Net.
They invented a layer "mask pooling" that they claim is better than ROI pooling and ROI align.
>As can be seen, our proposed
binary RoIMasking and ternary RoIMasking both outperform
RoIPool and RoIAlign in mAP0.7
. Specifically, our
ternary RoIMasking result improves the RoIAlign result by
around 2.5 points. This reflects that considering more context
information outside the proposals does help for salient
instance segmentation
Important benchmark attached:
https://i.imgur.com/wOF2Ovz.png