SEEmear: high-resolution georeferenced RGB-D data from apple orchard under palmette training system

This dataset contains high-resolution georeferenced RGB-D data collected from apple orchards using the SEEmear proximal sensing system. The data was acquired at Embrapa's Temperate Climate Fruit Growing Experimental Station (EFCT) in Vacaria, Rio Grande do Sul, Brazil, during field tests conducted in March 2025. Data Collection System: SEEmear integrates stereo cameras (ZED X, Stereolabs Inc.) with high-resolution global-shutter sensors (1920 × 1200 pixels), wide-angle lenses, and integrated IMUs, mounted on a mobile platform. The system includes RTK-GNSS positioning (Reach M2 rover and Reach RS3 base station) for centimeter-level accuracy, enabling precise georeferencing of all captured data. Dataset Contents: RGB-D stereo imagery at 30 fps IMU measurements synchronized with camera frames GNSS RTK positioning data GPS time synchronization for temporal consistency Orchard Characteristics: Apple cultivars: Fuji Training systems: palmette Camera-to-tree distance: 1.0-1.5 meters Data acquisition speed: approximately 5 km/h Applications: This dataset supports research in precision agriculture, plant phenotyping, and agricultural robotics, including: fruit detection and counting, 3D reconstruction and mapping, trunk and canopy structure analysis, SLAM algorithm development, and multi-view stereo reconstruction. Data Format: Complete RGB-D camera footage with associated IMU readings (SVO2 files) and georeferenced positioning data (POS), enabling accurate 3D point cloud generation and temporal mapping of orchard structures at sub-centimeter resolution. For reading data from SVO2, users should use the freely available Stereolabs ZED SDK.