OliveTreeSeg: A Diverse UAV-Based Dataset for Olive Tree Segmentation

OliveTreeSeg is a high-quality, manually curated image dataset designed to facilitate the development and evaluation of computer vision models for semantic segmentation of olive trees in agricultural settings. This dataset addresses key challenges in precision agriculture by providing a robust collection of 938 high-resolution RGB images (dimensions: 4608x3456 pixels) captured under real-world variability, enabling more generalizable models for tasks such as tree health monitoring, yield estimation, and automated orchard management. Acquired using a drone equipped with a Parrot Sequoia camera in a olive orchard in Mediterranean (Greece) during the growing seasons of 2024–2025, the dataset incorporates diverse tree sizes (both mature large-canopy and young small trees), varied viewing angles on tree grids, and a range of lighting conditions including morning sunlight, afternoon illumination, and cloudy skies. To support model training and benchmarking, the dataset is split into 655 training images, 141 validation images, and 142 test images, with pixel-level segmentation masks meticulously annotated by a domain expert for precise delineation of olive tree structures against background elements.