Dense Label Encoding for Boundary Discontinuity Free Rotation Detection

By Xue Yang, Liping Hou, Yue Zhou, Wentao Wang, Junchi Yan. Computer Vision and Pattern Recognition, 2020

298 citations recorded by the source corpus.

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 $\color{Red}{\text{DCL}}$ and is also integrated in our open source rotation detection benchmark:$\color{Red}{\text{RotationDetection}}$.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/729559024cc246c4bb7b33e395d2957b27662eb4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

1. Wentong Li
2. Qi Tian
3. Jianke Zhu
4. Junchi Yan

In Computer Vision and Pattern Recognition

1. Junchi Yan
2. Wentao Wang
3. Xue Yang
4. Jianke Zhu
5. Wentong Li
6. Qi Tian
7. Qi Ming
8. Xiaopeng Zhang

Citations in the corpus

• Learning High-Precision Bounding Box for Rotated Object Detection via Kullback-Leibler Divergence (2021, methodology): “Table 6 compares the six peer techniques, including IoU-Smooth L1 Loss [3], Modulated loss [44], RIL [33], CSL [4, 46], DCL [45], and GWD [5] on DOTA-v1.”
• Rethinking Rotated Object Detection with Gaussian Wasserstein Distance Loss (2021, methodology): “DCL [64] further solves square-like object detection problem introduced by the long edge definition, which refers to rotation insensitivity issue for instances that are approximately in square shape, which will be detailed in Sec.”
• Oriented RepPoints for Aerial Object Detection (2021, methodology): “[41] transform the angular regression to angular classification [40].”
• Poly Kernel Inception Network for Remote Sensing Detection (2024, background): “Consequently, numerous prior efforts have been dedicated to developing various oriented bounding box (OBB) detectors [10, 20, 31, 65, 67, 71] and improving the angle prediction accuracy for OBBs [68, 70, 72–74].”