Lightweight convolutional neural network for UAV onboard visual anomaly detection and localization
- 1State Key Laboratory of Spatial Datum
- 2RooXen Intellink Data Tech (Guangzhou) Co., Ltd
- 3School of Law and Business, Shaoyang University
Abstract
Unmanned aerial vehicle platforms increasingly demand efficient visual anomaly detection capabilities for infrastructure inspection and environmental monitoring applications. However, deploying sophisticated deep learning models on resource-constrained embedded processors remains challenging due to computational limitations. This paper proposes a lightweight convolutional neural network architecture specifically designed for UAV onboard anomaly detection and localization. The framework centers on a Ghost-Separable Block that integrates ghost convolution, depthwise separable convolution, and coordinate attention into a single unit tailored for aerial imagery; rather than a wholly new operator, it is a deliberate combination whose value we establish empirically, trimming parameters by 26% while preserving feature discrimination. The embedded coordinate attention preserves directional spatial information essential for accurate target localization in aerial imagery. The bidirectional feature pyramid network enables effective multi-scale fusion to handle scale variations inherent in UAV-captured scenes. A dual-branch prediction head with Complete Intersection over Union loss performs simultaneous anomaly classification and spatial regression. Extensive experiments on a collected UAV aerial anomaly dataset demonstrate that the proposed method achieves 81.2% mean average precision at 0.5 IoU threshold with only 2.18 million parameters, surpassing mainstream lightweight detectors such as YOLOv8n and EfficientDet-D0: it raises mean average precision while cutting parameters by 31% and computation by 59% relative to YOLOv8n. The dual-branch head and the Complete Intersection over Union loss are adopted rather than newly proposed; the block arrangement is where the contribution lies. Beyond the self-collected data, the model preserves its accuracy-efficiency advantage on the public VisDrone benchmark, and on a Jetson Xavier NX it sustains 42 FPS natively and 67 FPS after TensorRT FP16 optimization within a 15 W power envelope, confirming real-time onboard feasibility.
Related
- Comparative evaluation of convolutional and transformer-based object detection models for brain tumor localization in MRI
- Estimation and detection of rotor mass imbalance of wind turbines using convolutional neural network
- TCLNet: A Thermal Contrast-Aware Lightweight Network for infrared object detection
- Subject-independent Parkinson's Disease Detection through A Lightweight Spectrotemporal Hybrid Network.
- A convolutional autoencoder network based on multi-dimensional spatial feature reconstruction for anomaly detection of rotating machinery under complex operating conditions
BibTeX
@article{jiang2026lightweight,
title = {{Lightweight convolutional neural network for UAV onboard visual anomaly detection and localization}},
author = {Zihan Jiang and Guiping Zhao and Xiaoqi Xu and Peilong Zhang},
journal = {Scientific Reports},
year = {2026},
month = aug,
doi = {10.1038/s41598-026-68007-6},
url = {https://doi.org/10.1038/s41598-026-68007-6},
note = {CC BY},
}
RIS
TY - JOUR
TI - Lightweight convolutional neural network for UAV onboard visual anomaly detection and localization
AU - Zihan Jiang
AU - Guiping Zhao
AU - Xiaoqi Xu
AU - Peilong Zhang
PY - 2026
DA - 2026/08/01
JO - Scientific Reports
DO - 10.1038/s41598-026-68007-6
UR - https://doi.org/10.1038/s41598-026-68007-6
AB - Unmanned aerial vehicle platforms increasingly demand efficient visual anomaly detection capabilities for infrastructure inspection and environmental monitoring applications. However, deploying sophisticated deep learning models on resource-constrained embedded processors remains challenging due to computational limitations. This paper proposes a lightweight convolutional neural network architecture specifically designed for UAV onboard anomaly detection and localization. The framework centers on a Ghost-Separable Block that integrates ghost convolution, depthwise separable convolution, and coordinate attention into a single unit tailored for aerial imagery; rather than a wholly new operator, it is a deliberate combination whose value we establish empirically, trimming parameters by 26% while preserving feature discrimination. The embedded coordinate attention preserves directional spatial information essential for accurate target localization in aerial imagery. The bidirectional feature pyramid network enables effective multi-scale fusion to handle scale variations inherent in UAV-captured scenes. A dual-branch prediction head with Complete Intersection over Union loss performs simultaneous anomaly classification and spatial regression. Extensive experiments on a collected UAV aerial anomaly dataset demonstrate that the proposed method achieves 81.2% mean average precision at 0.5 IoU threshold with only 2.18 million parameters, surpassing mainstream lightweight detectors such as YOLOv8n and EfficientDet-D0: it raises mean average precision while cutting parameters by 31% and computation by 59% relative to YOLOv8n. The dual-branch head and the Complete Intersection over Union loss are adopted rather than newly proposed; the block arrangement is where the contribution lies. Beyond the self-collected data, the model preserves its accuracy-efficiency advantage on the public VisDrone benchmark, and on a Jetson Xavier NX it sustains 42 FPS natively and 67 FPS after TensorRT FP16 optimization within a 15 W power envelope, confirming real-time onboard feasibility.
SN - 2045-2322
ER -