/Detecting-defects-in-solar-panels

This project aims to detect defective solar panels in a solar array,using semantic segmentation of overhead IR image taken using drones.

Primary LanguageJupyter Notebook

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:

image image