Detecting-defects-in-solar-panels

Abstract

This project aims to detect defective solar panels in a solar array,by performing semantic segmentation. The panels are then segmented into

• Defect: LHS(Light Hot Spot)
• Defect: DHS(Dark Hot Spot)
• Defect: BP (ByPass diode)
• No defect: Normal

About Folders-Navigate easier:

• Final Notebook : Contains the finalized notebook which has all the models trained with information on errors and screenshots of predicted images.
• Annotation : Contains a text file about annotations used, steps taken and problems encountered.
• Preprocessing : Contains notebook used to generate the dataset and the details about steps taken during preprocessing , problems encountered and such.
• Training and results : Contains a summary of models trained and information on them.

Example of the defects: