Description:

This dataset includes high-resolution thermal images used to detect and diagnose issues in photovoltaic (PV) systems. Sourced from a research paper and a corresponding GitHub repository, it comprises 120 meticulously annotated thermal images. These annotations are comprehensive, catering to both instance segmentation and semantic segmentation tasks.

Download Dataset

Context

Photovoltaic (PV) systems are widely recognized for their ability to harness solar energy and convert it into electricity, contributing significantly to renewable energy solutions. However, like any technology, PV systems are not immune to failures. These failures often stem from various operational stresses such as thermal cycling, UV exposure, mechanical stresses, and electrical loading. Additionally, installation errors, including improper handling, poor connections, and substandard mounting, can further exacerbate the likelihood of failures.

Image Details

Thermal Images: Captured using FLIR cameras in R-JPG format, which includes radiometric data in the metadata.

Annotations: Extracted using the Flyr Python library, converting radiometric data to 8-bit images for segmentation tasks.

Importance

This dataset aids in the preventive maintenance of PV systems by enabling the identification of faulty modules through infrared thermography. Furthermore, when used with drones, it allows for extensive and efficient inspections. Consequently, maintenance teams can quickly pinpoint issues and address them before they escalate, ensuring optimal system performance.

This dataset is sourced from Kaggle.