{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/automatic-classification-of-defective","title":"Automatic Classification of Defective Photovoltaic Module Cells in Electroluminescence Images","arxiv_id":"1807.02894","date":"2018-07-08","proceeding":null,"authors":["Sergiu Deitsch","Vincent Christlein","Stephan Berger","Claudia Buerhop-Lutz","Andreas Maier","Florian Gallwitz","Christian Riess"],"abstract":"Electroluminescence (EL) imaging is a useful modality for the inspection of\nphotovoltaic (PV) modules. EL images provide high spatial resolution, which\nmakes it possible to detect even finest defects on the surface of PV modules.\nHowever, the analysis of EL images is typically a manual process that is\nexpensive, time-consuming, and requires expert knowledge of many different\ntypes of defects. In this work, we investigate two approaches for automatic\ndetection of such defects in a single image of a PV cell. The approaches differ\nin their hardware requirements, which are dictated by their respective\napplication scenarios. The more hardware-efficient approach is based on\nhand-crafted features that are classified in a Support Vector Machine (SVM). To\nobtain a strong performance, we investigate and compare various processing\nvariants. The more hardware-demanding approach uses an end-to-end deep\nConvolutional Neural Network (CNN) that runs on a Graphics Processing Unit\n(GPU). Both approaches are trained on 1,968 cells extracted from high\nresolution EL intensity images of mono- and polycrystalline PV modules. The CNN\nis more accurate, and reaches an average accuracy of 88.42%. The SVM achieves a\nslightly lower average accuracy of 82.44%, but can run on arbitrary hardware.\nBoth automated approaches make continuous, highly accurate monitoring of PV\ncells feasible.","url_abs":"http://arxiv.org/abs/1807.02894v3","url_pdf":"http://arxiv.org/pdf/1807.02894v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"anomaly-classification","task_name":"Anomaly Classification"},{"task_slug":null,"task_name":"GPU"}],"methods":[{"method_slug":"svm","method_name":"SVM"},{"method_slug":"vgg-19","method_name":"VGG-19"}],"datasets_introduced":[{"slug":"elpv","name":"ELPV","full_name":"A dataset of functional and defective solar cells extracted from EL images of solar modules"}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1807.02894","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}