Multi-temporal UAV RGB image datasets for individual tree segmentation

This dataset comprises UAV-acquired RGB image samples covering three distinct forest ecosystem types across multiple phenological seasons. Each dataset package contains high-resolution PNG-format aerial imagery paired with corresponding annotation files, maintaining consistent filenames between images and their pixel-level vegetation labels for seamless data association. The time-series acquisition strategy captures seasonal variations in canopy structure, coloration, and density within identical geographic coordinates. Researchers can perform cross-seasonal comparative analysis by pairing summer and autumn/winter images from the same forest stand, enabling multi-temporal object detection of individual tree crowns with changing environmental conditions. This resource facilitates algorithm development for precision forestry, ecological monitoring, and climate change impact studies, particularly benefiting models requiring temporal feature learning from limited spectral (RGB) data. The standardized structure ensures compatibility with mainstream deep learning frameworks for vegetation remote sensing applications.