EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ATDIOU: Arctangent Differential Loss Function for Bounding Box Regression

Qiang Tang, Hao Qiang iD, Yuan Tian, Xubin Feng iD, Wei Hao, Meilin Xie

Sensors · 01.03.2026 · 26 (5) · 1545

DOI: 10.3390/s26051545 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 01.03.2026. Verfasst von Qiang Tang und 5 weiteren Autor:innen. Beteiligte Einrichtung: Xi'an Institute of Optics and Precision Mechanics of CAS, Xi'an 710119, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Xi'an Institute of Optics and Precision Mechanics of CAS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Object detection is a fundamental task in computer vision. Bounding box regression (BBR) losses are critical to detector performance. However, evaluation measures that rely on the Intersection over Union (IoU) between the predicted and ground truth boxes are highly sensitive to positional deviations, which can hinder optimization. To alleviate this issue, we propose ATDIoU, a novel arctangent-differential loss for bounding-box regression. ATDIoU computes distance similarity between a predicted and a ground truth box by modeling the distances between their corresponding vertices as a two-dimensional arctangent differential distribution (ATD). This arctangent differential-based design mitigates bounding box drift and reduces sensitivity to localization errors. As a result, it guides the model to learn target positions more effectively. We evaluate ATDIoU by integrating it into YOLOv6 and conducting experiments on PASCAL VOC and VisDrone2019. The results demonstrate that ATDIoU yields improvements of 1.4% and 0.7% in mean average precision (mAP) relative to MPDIoU.

Schlagwörter

Stichwörter der Autor:innen

arctangent-differential functionbounding box regressioncomputer visionobject detection

Autor:innen und Institutionen

  • Xi'an Institute of Optics and Precision Mechanics of CAS, Xi'an 710119, China.
  • University of Chinese Academy of Sciences, Beijing 100049, China.

ORCID-Profile: Hao Qiang, Xubin Feng

Förderung

  • Xi'an Institute of Optics and Precision Mechanics of CAS

Zitieren

APA

Tang, Q., Qiang, H., Tian, Y., Feng, X., Hao, W., & Xie, M. (2026). ATDIOU: Arctangent Differential Loss Function for Bounding Box Regression. Sensors, 26(5), 1545. https://doi.org/10.3390/s26051545

Vancouver

Tang Q, Qiang H, Tian Y, Feng X, Hao W, Xie M. ATDIOU: Arctangent Differential Loss Function for Bounding Box Regression. Sensors. 2026;26(5):1545. doi:10.3390/s26051545

BibTeX
@article{Tang2026,
  author = {Tang, Qiang and Qiang, Hao and Tian, Yuan and Feng, Xubin and Hao, Wei and Xie, Meilin},
  title = {ATDIOU: Arctangent Differential Loss Function for Bounding Box Regression},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {5},
  pages = {1545},
  doi = {10.3390/s26051545}
}
Kostenlos starten →