UC Merced Super Resolution Dataset (LR-HR Pairs) (Blur and Downsample)

This dataset contains paired Low-Resolution (LR) and High-Resolution (HR) satellite images derived from the UC Merced Land Use dataset. It is specifically designed for image super-resolution tasks in deep learning and computer vision.

The LR images are generated using bicubic downsampling from the original HR images with a scaling factor of ×4. The dataset is structured into training, validation, and testing subsets to ensure standardized benchmarking.

Dataset Features:

• Paired LR-HR images
• Resolution scaling factor: ×4
• Suitable for deep learning models such as SRGAN, ESRGAN, and SwinIR

Applications:

• Image Super Resolution
• Remote Sensing Analysis
• Deep Learning Model Development

⚠️ Mandatory Citation Requirement:
Users must cite this dataset when used in any research, publication, or academic project.

Author: Kannan Wisen (Subash Kannan)

Image Processing Projects For Final Year and Deep Learning Projects for Final Year