Semantic Segmentation of Boats, Landscapes, and Natural Elements
Worked on a semantic segmentation project involving outdoor scenes containing boats, sea, sky, vegetation, and land. Using CVAT, I accurately segmented all objects according to the dataset instructions: Segmented all boats with pixel-level precision. Annotated the sky, sea, and land (including sand and bridges). Segmented trees, plants, and all greenery as vegetation. Labeled any remaining objects without a specific class as “Other.” Followed strict semantic segmentation rules and ensured high pixel accuracy across all classes. Reviewed masks for consistency, corrected overlaps, and ensured class purity within each region. This project strengthened my ability to follow detailed guidelines, maintain annotation accuracy, and perform high-quality semantic segmentation on complex images.