Voltmeter defect detection is a critical component in the calibration pipeline of electric meters. Traditional calibration methods heavily rely on manual inspection, resulting in time-consuming procedures with high rates of misjudgment. The scarcity of authentic samples containing defects poses challenges in constructing a sufficiently abundant dataset for defect samples. However, current Deformable DETR demonstrates suboptimal performance on small-scale datasets. One factor contributing to its underperformance on small-scale datasets is the excessive redundancy present in queries within the encoder, posing challenges for the model to effectively concentrate on objects. Moreover, the Hungarian matching in Deformable DETR results in a scarcity of positive examples, which hampers convergence speed. This paper introduces enhancements to Deformable DETR, including a sparse encoder and a hybrid matching mechanism, aimed at resolving the slow convergence problem on small-scale datasets. Finally, extensive experiments on our dataset validate the superior effectiveness of our proposed method, achieving the best performance in defect detection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.