Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Dense Label Encoding for Boundary Discontinuity Free Rotation Detection

About

Rotation detection serves as a fundamental building block in many visual applications involving aerial image, scene text, and face etc. Differing from the dominant regression-based approaches for orientation estimation, this paper explores a relatively less-studied methodology based on classification. The hope is to inherently dismiss the boundary discontinuity issue as encountered by the regression-based detectors. We propose new techniques to push its frontier in two aspects: i) new encoding mechanism: the design of two Densely Coded Labels (DCL) for angle classification, to replace the Sparsely Coded Label (SCL) in existing classification-based detectors, leading to three times training speed increase as empirically observed across benchmarks, further with notable improvement in detection accuracy; ii) loss re-weighting: we propose Angle Distance and Aspect Ratio Sensitive Weighting (ADARSW), which improves the detection accuracy especially for square-like objects, by making DCL-based detectors sensitive to angular distance and object's aspect ratio. Extensive experiments and visual analysis on large-scale public datasets for aerial images i.e. DOTA, UCAS-AOD, HRSC2016, as well as scene text dataset ICDAR2015 and MLT, show the effectiveness of our approach. The source code is available at https://github.com/Thinklab-SJTU/DCL_RetinaNet_Tensorflow and is also integrated in our open source rotation detection benchmark: https://github.com/yangxue0827/RotationDetection.

Xue Yang, Liping Hou, Yue Zhou, Wentao Wang, Junchi Yan• 2020

Related benchmarks

TaskDatasetResultRank
Oriented Object DetectionDOTA v1.0 (test)
SV79.04
378
Object DetectionDOTA 1.0 (test)
Plane AP89.26
256
Oriented Object DetectionDOTA (test)
AP (Plane)89.26
92
Object DetectionHRSC 2016 (test)
mAP@0.0789.46
72
Oriented Object DetectionDOTA v1.5 (test)--
58
Oriented Object DetectionHRSC 2016 (test)
mAP89.46
55
Object DetectionUCAS-AOD (test)--
27
Oriented Object DetectionDOTA single-scale 1.0 (test)
AP (PL)89.78
24
Rotation Object DetectionDOTA v2.0 (test)
AP5048.43
14
Rotation Object DetectionDOTA v1.0 (val)
AP5070.66
14
Showing 10 of 12 rows

Other info

Code

Follow for update