THAB Equirectangular Dataset

This dataset contains 4,750 paired equirectangular RGB and depth (range) images collected in Aschaffenburg, Germany using a fully synchronized sensor rig. RGB images were captured using a Teledyne FLIR Ladybug 5+ omnidirectional camera. Depth images were recorded using an Ouster OS2-128 Rev 7 LiDAR and projected into the same equirectangular format. The dataset is organized into 9 sequences, each representing a distinct collection run in various parts of Aschaffenburg. Key Features High-resolution 360° equirectangular color images. Corresponding equirectangular depth maps aligned with the RGB images. Time-synchronized multimodal data from camera and LiDAR sensors. Diverse urban and suburban environments. Privacy compliant: All faces and license plates in RGB images have been anonymized. Linked to SemanticTHAB: This dataset is temporally and spatially aligned with the SemanticTHAB dataset containing semantically labeled point clouds. Python mapping script provided to project semantic labels from SemanticTHAB’s point clouds onto the RGB and depth images. Potential Applications 3D scene reconstruction Sensor fusion research Panoramic depth estimation Autonomous driving and robotics Semantic scene understanding