LightR-YOLOv5: A compact rotating detector for SARS-CoV-2 antigen-detection rapid diagnostic test results

• Published in: Displays Vol. 78; p. 102403
• Main Authors: Wang, Rongsheng, Duan, Yaofei, Hu, Menghan, Liu, Xiaohong, Li, Yukun, Gao, Qinquan, Tong, Tong, Tan, Tao
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.07.2023
• Subjects: Lightweight; RDT detection; Rotating object detection; YOLOv5; Rotating object detection; YOLOv5; Lightweight; RDT detection
• ISSN: 0141-9382; 1872-7387; 1872-7387
• DOI: 10.1016/j.displa.2023.102403

Nucleic acid testing is currently the golden reference for coronaviruses (SARS-CoV-2) detection, while the SARS-CoV-2 antigen-detection rapid diagnostic tests (RDT) is an important adjunct. RDT can be widely used in the community or regional screening management as self-test tools and may need to be verified by healthcare authorities. However, manual verification of RDT results is a time-consuming task, and existing object detection algorithms usually suffer from high model complexity and computational effort, making them difficult to deploy. We propose LightR-YOLOv5, a compact rotating SARS-CoV-2 antigen-detection RDT results detector. Firstly, we employ an extremely light-weight L-ShuffleNetV2 network as a feature extraction network with a slight reduction in recognition accuracy. Secondly, we combine semantic and texture features in different layers by judiciously combining and employing GSConv, depth-wise convolution, and other modules, and further employ the NAM attention to locate the RDT result detection region. Furthermore, we propose a new data augmentation approach, Single-Copy–Paste, for increasing data samples for the specific task of RDT result detection while achieving a small improvement in model accuracy. Compared with some mainstream rotating object detection networks, the model size of our LightR-YOLOv5 is only 2.03MB, and it is 12.6%, 6.4%, and 7.3% higher in mAP@.5:.95 metrics compared to RetianNet, FCOS, and R3Det, respectively. •We are the first to propose a framework to deliver accurate RDT results detection with fast responses. The solution reduces the workload for human detection which could be beneficial to smart medical care.•Propose an improved lightweight feature extraction network, L-ShuffleNetV2, which can be widely used in other detection tasks. Design a lightweight feature fusion module and add the parameter-free NAM attention to obtain faster and higher precision detection results.•Propose a novel data augmentation, Single-Copy–Paste. This method is to copy and paste the object on one single image. Compared with the traditional Copy–Paste, our proposed method can achieve data augmentation without changing the data distribution.•Propose a lightweight rotating object detection to address the variable result area in RDT result detection. This method can effectively detect the geometric contour and position information of the result and reduce the redundancy of information.