Loading…
Directional Edge Boxes: Exploiting Inner Normal Direction Cues forEffective Obiect Prooosal Generation
Edges are important cues for localizing object proposals. The recent progresses to this problem are mostlydriven by defining effective objectness measures based on edge cues. In this paper, we develop a new representation nameddirectional edges on which each edge pixel is assigned with a direction t...
Saved in:
Published in: | 计算机科学技术学报:英文版 2017, Vol.32 (4), p.701-713 |
---|---|
Main Author: | |
Format: | Article |
Language: | English |
Subjects: | |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Edges are important cues for localizing object proposals. The recent progresses to this problem are mostlydriven by defining effective objectness measures based on edge cues. In this paper, we develop a new representation nameddirectional edges on which each edge pixel is assigned with a direction toward object center, through learning a directionprediction model with convolutional neural networks in a holistic manner. Based on directional edges, two new objectnessmeasures are designed for ranking object proposals. Experiments show that the proposed method achieves 97.1% objectrecall at an overlap threshold of 0.5 and 81.9% object recall at an overlap threshold of 0.7 at 1 000 proposals on the PASCALVOC 2007 test dataset, which is suDerior to the state-of-the-art methods. |
---|---|
ISSN: | 1000-9000 1860-4749 |