Defect Detection of Photovoltaic Panels to Suppress Endogenous Shift Phenomenon

Abstract

Efficient and intelligent surface defect detection of photovoltaic modules is crucial for improving the quality of photovoltaic modules and ensuring the reliable operation of large-scale infrastructure. However, the scenario characteristics of data distribution deviation make the construction of defect detection models for open world scenarios such as photovoltaic manufacturing and power plant inspections a challenge. Therefore, we propose the Gather and Distribute Domain shift Suppression Network. It adopts a single domain generalized method that is completely independent of the test samples to address the problem of distribution shift. Using a one-stage network as the baseline network breaks through the limitations of traditional domain generalization methods that typically use two-stage networks. It not only balances detection accuracy and speed but also simplifies the model deployment and application process. The network first employs the DeepSpine module to capture a wider range of contextual information. By concatenating and aligning multi-scale channel features, it effectively suppresses background style shifts. Building upon this, the Gather and Distribute Module performs cross layer interactive learning on multi-scale channel features. The multi-level features and semantic dependencies learned enhance the localization and recognition ability of target defects, thereby achieving the suppression of defect instance shift. Furthermore, we utilizes normalized Wasserstein distance for similarity measurement, reducing measurement errors caused by bounding box position deviations. We conducted a comprehensive evaluation of our network on the Electroluminescence Endogenous Shift Dataset and Photovoltaic Inspection Infrared Dataset. In scenarios with three production lines and four heights on two datasets, the detection accuracy of GDDS reached 91.2%, 82.3%, 79.9%, and 92.8%, 82.7%, 77.2%, and 69.2%, respectively. The experimental results showed that our method can adapt to defect detection in open world scenarios faster and better than other state-of-the-art methods.