Sample frames for this project are available under NDA. Write to bilal@wortel.ai and we will share the relevant before and after pairs.
Challenges
- Identifying target trees in complex outdoor environments.
- Differentiating catch trees from surrounding trees and vegetation.
- Overlapping tree canopies were hard to separate cleanly.
- Partially visible trees required judgement about mask extent.
- Dense branches and foliage complicated boundary tracing.
- Complex backgrounds contained buildings, roads, crops and other objects.
- Camera angles, lighting conditions and tree sizes varied throughout the dataset.
- Maintaining precise segmentation boundaries around irregular tree canopies.
Related work
Aerial Orthomosaic Segmentation
Whole-site orthomosaics segmented into vegetation, water and bare ground across tiles stitched from multiple flights.
Ground Cover Quadrat Segmentation
2,275 cover-crop plot images annotated in Labelbox: plot corners boxed, then every pixel classed as groundcover, weed, residue, crop or soil.
Black Grass Segmentation
Polygon masks isolating black grass from crops and similar-looking grasses across dense agricultural field imagery.