PGWave-RotNe: A Novel Lightweight Network for Oriented Object Detection in Remote Sensing.
Saved in:
| Title: | PGWave-RotNe: A Novel Lightweight Network for Oriented Object Detection in Remote Sensing. |
|---|---|
| Authors: | Wang, Donglong1 (AUTHOR), Cao, Tieyong1 (AUTHOR) cty_ice@aeu.edu.cn, Yang, Jibin1 (AUTHOR), SongGong, Kunkun1 (AUTHOR) |
| Source: | Remote Sensing. Jun2026, Vol. 18 Issue 11, p1760. 27p. |
| Subjects: | Remote sensing, Object recognition (Computer vision), Loss functions (Statistics), Feature extraction |
| Abstract: | Highlights: What are the main findings? This work introduces PGWave-RotNet, an effective lightweight integration scheme specifically designed for oriented object detection in remote sensing imagery. The framework comprises four key components: Partial-Ghost Shuffle Convolution (PGSConv), Gated Position-Sensitive Attention (GPSA), Directional-Biased Wavelet Transform Upsampling (DBWTU), and a Weighted Cosine Angular Loss (WCAL). Together, these modules form a lightweight pipeline that enables accurate oriented bounding box prediction across diverse object categories in remote sensing scenes. Extensive validation is performed on two public benchmarks, DOTAv1 and DIOR-R, using comparative, ablation, and generalization experiments. Visual and quantitative results indicate that the proposed model consistently outperforms baseline detectors, with particularly notable gains when handling objects that exhibit large scale variations and arbitrary orientations. What are the implications of the main findings? The results indicate that customizing architectural enhancements to the inherent properties of remote sensing objects—namely, pronounced scale disparities, orientation heterogeneity, and a high prevalence of small targets—can substantially elevate detection performance while maintaining an advantageous trade-off between model compactness and computational cost. Furthermore, the proposed PGWave-RotNet demonstrates a strong combination of low parameter count and high inference speed, validating its practicality for real-world remote sensing object detection scenarios, especially those requiring deployment on resource-constrained platforms. Accurate and efficient oriented detection is critical for remote sensing images, yet remains challenging due to multi-scale distribution, arbitrary orientations, and stringent computational constraints of onboard platforms. To mitigate these challenges, we propose Partial-Ghost Shuffle Convolution and Gated Position-Sensitive Attention Wavelet Rotation Network (PGWave-RotNet), a lightweight wavelet-guided rotation detector that explicitly enhances multi-scale and arbitrarily oriented features while maintaining high efficiency. To reduce feature redundancy while preserving directional diversity, we design a Partial-Ghost Shuffle Convolution (PGSConv) module that integrates partial convolution with ghost shuffle. Next, to adaptively refine multi-scale and arbitrarily oriented contexts, we introduce a Gated Position-Sensitive Attention (GPSA) module with a learnable gating mechanism. To suppress aliasing and sharpen edges during upsampling, we propose a Directional-Biased Wavelet Transform Upsampling (DBWTU) module based on high-frequency wavelet reconstruction. Additionally, we develop a Weighted Cosine Angular Loss (WCAL) to improve orientation precision for square-like targets. Experiments on DOTAv1 and DIOR-R achieve 82.27% and 83.82% mAP50, outperforming existing methods. These innovations collectively enable efficient and accurate oriented detection in remote sensing. [ABSTRACT FROM AUTHOR] |
| Copyright of Remote Sensing is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Highlights: What are the main findings? This work introduces PGWave-RotNet, an effective lightweight integration scheme specifically designed for oriented object detection in remote sensing imagery. The framework comprises four key components: Partial-Ghost Shuffle Convolution (PGSConv), Gated Position-Sensitive Attention (GPSA), Directional-Biased Wavelet Transform Upsampling (DBWTU), and a Weighted Cosine Angular Loss (WCAL). Together, these modules form a lightweight pipeline that enables accurate oriented bounding box prediction across diverse object categories in remote sensing scenes. Extensive validation is performed on two public benchmarks, DOTAv1 and DIOR-R, using comparative, ablation, and generalization experiments. Visual and quantitative results indicate that the proposed model consistently outperforms baseline detectors, with particularly notable gains when handling objects that exhibit large scale variations and arbitrary orientations. What are the implications of the main findings? The results indicate that customizing architectural enhancements to the inherent properties of remote sensing objects—namely, pronounced scale disparities, orientation heterogeneity, and a high prevalence of small targets—can substantially elevate detection performance while maintaining an advantageous trade-off between model compactness and computational cost. Furthermore, the proposed PGWave-RotNet demonstrates a strong combination of low parameter count and high inference speed, validating its practicality for real-world remote sensing object detection scenarios, especially those requiring deployment on resource-constrained platforms. Accurate and efficient oriented detection is critical for remote sensing images, yet remains challenging due to multi-scale distribution, arbitrary orientations, and stringent computational constraints of onboard platforms. To mitigate these challenges, we propose Partial-Ghost Shuffle Convolution and Gated Position-Sensitive Attention Wavelet Rotation Network (PGWave-RotNet), a lightweight wavelet-guided rotation detector that explicitly enhances multi-scale and arbitrarily oriented features while maintaining high efficiency. To reduce feature redundancy while preserving directional diversity, we design a Partial-Ghost Shuffle Convolution (PGSConv) module that integrates partial convolution with ghost shuffle. Next, to adaptively refine multi-scale and arbitrarily oriented contexts, we introduce a Gated Position-Sensitive Attention (GPSA) module with a learnable gating mechanism. To suppress aliasing and sharpen edges during upsampling, we propose a Directional-Biased Wavelet Transform Upsampling (DBWTU) module based on high-frequency wavelet reconstruction. Additionally, we develop a Weighted Cosine Angular Loss (WCAL) to improve orientation precision for square-like targets. Experiments on DOTAv1 and DIOR-R achieve 82.27% and 83.82% mAP50, outperforming existing methods. These innovations collectively enable efficient and accurate oriented detection in remote sensing. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 20724292 |
| DOI: | 10.3390/rs18111760 |
