Used retrained pv-vision (original: https://github.com/hackingmaterials/pv-vision, retrained: https://github.com/sandialabs/pvcracks/tree/main/retrain) to extract crack masks from module EL images (uploaded here). These images are used to train our variational autoencoder (VAE) for feature bundling (in the latent space) and crack mask generation (https://github.com/sandialabs/pvcracks/tree/main/VAE)

Dataset Metadata

מחבר normanj
דוא"ל התחזוקה nrjost@sandia.gov
Institution SNL
Data Source Type Lab and Experimental
Accelerated? No
Environmental Conditions Not Applicable
Measurement Types EL
Owner normanj

מחבר normanj
Updated יולי 3, 2024, 23:19 (UTC)
נוצר ב יולי 3, 2024, 23:18 (UTC)