Improved SSD-assisted algorithm for surface defect detection of electromagnetic luminescence

Abstract

Defect detection of electromagnetic luminescence (EL) cells is the core step in the production and preparation of solar cell modules to ensure conversion efficiency and long service life of batteries. However, due to the lack of feature extraction capability for small feature defects, the traditional single shot multibox detector (SSD) algorithm performs not well in EL defect detection with high accuracy. Consequently, an improved SSD algorithm with modification in feature fusion in the framework of deep learning is proposed to improve the recognition rate of EL multi-class defects. A dataset containing images with four different types of defects through rotation, denoising, and binarization is established for the EL. The proposed algorithm can greatly improve the detection accuracy of the small-scale defect with the idea of feature pyramid networks. An experimental study on the detection of the EL defects shows the effectiveness of the proposed algorithm. Moreover, a comparison study shows the proposed method outperforms other traditional detection methods, such as the SIFT, Faster R-CNN, and YOLOv3, in detecting the EL defect.

Detection algorithm for bearing roller end surface defects based on improved YOLOv5n and image fusion

“…The gamma-corrected pixel value I(i, j) can be calculated from equation (11) as I(i, j) = ((I(i, j) + 0.5)/256) 1 gamma * 256 − 0.5 (11) where I(i, j) denotes each pixel value and gamma takes a value in the range 0.05-5.…”

“…Xu et al [10], for example, proposed an improved Faster-RCNN to improve the ability to detect sand inclusion defects. An improved SSD algorithm was used to detect defects in electromagnetic luminescence cells, which greatly improved the accuracy of detection of small-scale defects [11]. In addition, Jing et al [12] used an improved YOLOv3 for fabric defect detection and their experimental results showed that the error detection rate of the improved network model was less than 5% for both gray cloth and checked cloth, giving it good practical application value.…”

ESDDNet: efficient small defect detection network of workpiece surface

“…Liu et al proposed the single-shot detector (SSD) [18], which added auxiliary structures on the basis of the Visual Geometry Group (VGG-16) [19] to improve performance. Xu etal [20] proposed an improved SSD-assisted algorithm for electromagnetic luminescence surface defect detection, which can improve the recognition rate of multi-class defects. Lin etal proposed RetinaNet and Focal Loss [21], which effectively balanced positive and negative samples and achieved more effective training.…”