Weakly Supervised Segmentation of Cracks on Solar Cells Using Normalized L_p Norm

Abstract

Photovoltaic is one of the most important renewable energy sources for dealing with world-wide steadily increasing energy consumption. This raises the demand for fast and scalable automatic quality management during production and operation. However, the detection and segmentation of cracks on electroluminescence (EL) images of mono-or polycrystalline solar modules is a challenging task. In this work, we propose a weakly supervised learning strategy that only uses image-level annotations to obtain a method that is capable of segmenting cracks on EL images of solar cells. We use a modified ResNet-50 to derive a segmentation from network activation maps. We use defect classification as a surrogate task to train the network. To this end, we apply normalized L p normalization to aggregate the activation maps into single scores for classification. In addition, we provide a study how different parameterizations of the normalized L p layer affect the segmentation performance. This approach shows promising results for the given task. However, we think that the method has the potential to solve other weakly supervised segmentation problems as well.

Deep‐learning‐based pipeline for module power prediction from electroluminescense measurements

“…Although the choice of a ResNet18 and the reported hyperparameters are specific to the dataset used in this work, we are confident that they hold for other datasets as well, since previous experiments on defect recognition lead to similar conclusions. 13 For this work, we use a dataset of 719 EL measurements including three module types. However, we experimentally show that the model does not generalize well to unseen module types.…”

“…In a recent work, we used this to segment cracks in EL measurements using a ResNet18. 13 We propose to build upon this work and use a slightly modified ResNet18 that allows one to infer the predicted power loss for an arbitrary area by integrating over the respective area of the CAM. The…”

Anomaly Detection and Automatic Labeling for Solar Cell Quality Inspection Based on Generative Adversarial Network

“…Or we explicitly train a Fully Convolutional Network to perform pixel-wise classification [ 29 ]. Additionally, other researchers have also proposed other types of defect location, using bounding boxes [ 30 , 31 ], or by visualizing the activation maps from the last network layer [ 25 , 32 ].…”

State of AI-Based Monitoring in Smart Manufacturing and Introduction to Focused Section

“…A variety of benchmark datasets are available online for QI to verify the developed inspection algorithms and to compare different methods, including road crack datasets [56], datasets for PCB analysis [57], nanofibrous materials datasets [58], steel strip surface datasets [59], X-Ray datasets [60], saliency defects of magnetic tile [61], images of cracks on solar cells [62], non-woven fabric [62], etc.…”