Thermal Imaging Dataset for Hotspot Detection on Solar Panels: Impact of Bird Droppings on Efficiency

Dataset Description:
This dataset consists of thermal images of solar panels captured using a UTi165A thermal imaging camera. It has been developed and annotated by the Tutankhamun Research School for Artificial Intelligence (TRSAI), Egypt, https://trsai.org. The primary objective is to identify and analyze hot spots on solar panels, which are often caused by external factors such as bird droppings. These hotspots can reduce panel efficiency and lead to localized overheating, making this dataset a valuable resource for maintenance and efficiency analysis.

The dataset showcases TRSAI's dedication to advancing artificial intelligence applications in sustainable energy solutions. Authored under the guidance of TRSAI, this work reflects the school’s commitment to integrating AI with real-world environmental challenges.

Camera Specifications:
- Resolution: 160 × 120 pixels
- Thermal Sensitivity (NETD): ≤ 50 mK
- Spatial Resolution (IFOV): 6.1 mrad
- Frame Rate: ≤ 9 Hz
- Spectral Range: 8~14 µm
- Accuracy: ±2℃ or ±2%, whichever is greater
- Emissivity: Adjustable between 0.01 to 1.00 (default: 0.95)

Dataset Versions:
The dataset is divided into two versions:

Version 1:
- Total Images: 410
- Training Set: 304 images (74%)
- Validation Set: 70 images (17%)
- Test Set: 36 images (9%)
- Preprocessing:
- Auto-Orientation applied.
- Images resized to 640 × 640 pixels.
- 15% of images converted to grayscale.
- Augmentation Techniques:
- Horizontal and vertical flipping.
- Rotation (90° clockwise, counter-clockwise, and upside down).

Version 2:
- Total Images: 850
- Training Set: 744 images (88%)
- Validation Set: 70 images (8%)
- Test Set: 36 images (4%)
- Preprocessing:
- Auto-Orientation applied.
- Images resized to 640 × 640 pixels.
- Grayscale applied.
- Augmentation Techniques:
- Horizontal and vertical flipping.
- Rotation (90° clockwise, counter-clockwise, and upside down).
- Brightness adjustments (-20% to +20%).
- Exposure adjustments (-10% to +10%).

Objective:
This dataset is designed to aid in the detection of thermal anomalies, particularly hot spots resulting from bird droppings. These anomalies can negatively impact the efficiency of solar panels, making it essential for maintenance teams to address such issues promptly. The dataset supports research in thermal imaging and AI-based solutions for sustainable energy management.