EUVIMEDEUVIMED entdecken
Crossref

Lightweight Monocular Depth Estimation Based on Structural Fusion and Edge-Aware Distillation

Dongliang Wang, Zongsheng Zhou, Ming Jin, Qinghua Guo, Yeyao Chen, Xiangqian Fang

IEEE Sensors Journal · 2026-11 · 26 (21) · 33198-33211

DOI: 10.1109/jsen.2026.3736063 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-11. Verfasst von Dongliang Wang und 5 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China, Science and Technology Innovation 2035 Major Project of Ningbo.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Self-supervised monocular depth estimation (MDE) eliminates the reliance on expensive ground-truth depth annotations and has emerged as a crucial technique for 3-D scene understanding in autonomous driving and robotic navigation. In this article, a lightweight self-supervised MDE model is proposed for balancing global context modeling and computational efficiency through structural fusion and edge-aware distillation. The model integrates an efficient model (EMO) as its backbone, effectively combining the local feature extraction capability of CNNs with the global reasoning ability of Transformers. To enhance feature representation in the decoder, a spatial-channel fusion attention (SC-FA) module is proposed, which improves cross-level feature fusion while preserving spatial relationships essential for depth coherence. Furthermore, an edge-aware knowledge distillation framework is established to transfer geometric priors from a pretrained MonoViT teacher model, enabling model compression without sacrificing accuracy. Experiments on the KITTI benchmark demonstrate the competitive performance of our method: our full model achieves an absolute relative error (Abs Rel) of 0.104 and a threshold accuracy of 0.893 with only 5.8 M parameters and $\delta \lt {1}.{25}$ , outperforming Monodepth2-Res18 (14.3 M, 0.115 Abs Rel, 0.883 accuracy with $\delta \lt {1}.{25}$ ). Furthermore, its scalable lightweight variants enable flexible complexity reduction, making it adaptable for real-time deployment on resource-constrained edge devices.

Förderung

  • National Natural Science Foundation of China
  • Zhejiang Provincial Natural Science Foundation of China
  • Science and Technology Innovation 2035 Major Project of Ningbo

Zitieren

APA

Wang, D., Zhou, Z., Jin, M., Guo, Q., Chen, Y., & Fang, X. (2026). Lightweight Monocular Depth Estimation Based on Structural Fusion and Edge-Aware Distillation. IEEE Sensors Journal, 26(21), 33198-33211. https://doi.org/10.1109/jsen.2026.3736063

Vancouver

Wang D, Zhou Z, Jin M, Guo Q, Chen Y, Fang X. Lightweight Monocular Depth Estimation Based on Structural Fusion and Edge-Aware Distillation. IEEE Sensors Journal. 2026;26(21):33198-33211. doi:10.1109/jsen.2026.3736063

BibTeX
@article{Wang2026,
  author = {Wang, Dongliang and Zhou, Zongsheng and Jin, Ming and Guo, Qinghua and Chen, Yeyao and Fang, Xiangqian},
  title = {Lightweight Monocular Depth Estimation Based on Structural Fusion and Edge-Aware Distillation},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {21},
  pages = {33198--33211},
  doi = {10.1109/jsen.2026.3736063}
}
Kostenlos starten →