We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CV

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: EOLO: Embedded Object Segmentation only Look Once

Abstract: In this paper, we introduce an anchor-free and single-shot instance segmentation method, which is conceptually simple with 3 independent branches, fully convolutional and can be used by easily embedding it into mobile and embedded devices.
Our method, refer as EOLO, reformulates the instance segmentation problem as predicting semantic segmentation and distinguishing overlapping objects problem, through instance center classification and 4D distance regression on each pixel. Moreover, we propose one effective loss function to deal with sampling a high-quality center of gravity examples and optimization for 4D distance regression, which can significantly improve the mAP performance. Without any bells and whistles, EOLO achieves 27.7$\%$ in mask mAP under IoU50 and reaches 30 FPS on 1080Ti GPU, with a single-model and single-scale training/testing on the challenging COCO2017 dataset.
For the first time, we show the different comprehension of instance segmentation in recent methods, in terms of both up-bottom, down-up, and direct-predict paradigms. Then we illustrate our model and present related experiments and results. We hope that the proposed EOLO framework can serve as a fundamental baseline for a single-shot instance segmentation task in Real-time Industrial Scenarios.
Comments: 7 pages, 5 figures, 2 tables, 25 conferences
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
MSC classes: 68T07
ACM classes: I.4.6
Cite as: arXiv:2004.00123 [cs.CV]
  (or arXiv:2004.00123v1 [cs.CV] for this version)

Submission history

From: Longfei Zeng [view email]
[v1] Tue, 31 Mar 2020 21:22:05 GMT (581kb,D)

Link back to: arXiv, form interface, contact.