2,100 RGB images of urban locations across 21 land use classes, with exactly 100 images per category. Each image is 256x256 pixels with a 1-foot resolution, extracted from the USGS National Map Urban Area Imagery collection.
Use Cases
- Train image classification models to identify specific land use types using the 21 provided labels
- Evaluate the performance of deep learning architectures on high-resolution 256x256 RGB aerial imagery
- Perform transfer learning experiments using the balanced 100-image-per-class distribution
Strengths
- 2,100 RGB images with a spatial resolution of 1 foot per pixel
- 21 distinct land use classes with exactly 100 images per class
- Standardized 256x256 pixel image dimensions for consistent model input
- Data sourced from the USGS National Map Urban Area Imagery collection