Content
The dataset includes two versions of each image: a colorized and an annotated ground-truth version. Each image measures 288×156 pixels, depicting four distinct classes:
- High turbidity level
- Moderate turbidity level
- Low turbidity level
- Presence of aquatic plants
Applications
This dataset is valuable for exploring land use classification in remote sensing, particularly in monitoring water quality and sediment yields in hydroelectric reservoirs. Researchers can use it to assess the efficacy of traditional classifiers and explore modern deep learning techniques for improving remote sensing tasks.
Use Cases and Inspirations
- Enhancing hydroelectric reservoir management
- Improving sediment classification in water bodies
- Bridging the gap in remote sensing datasets, particularly from Brazilian hydroelectric facilities
- Applying computer vision for better land cover and water quality analysis