UAVStereo: A Multiple Resolution Dataset for Stereo Matching in UAV Scenarios

Stereo matching is a fundamental task in three-dimensional scene reconstruction. Recently, deep learning-based methods have proven effective on some benchmark datasets, such as KITTI and SceneFlow. Unmanned aerial vehicles (UAVs) are commonly used for surface observation, the images captured frequently detailed 3D reconstruction because of their high resolution low-altitude acquisition. Currently, mainstream supervised learning networks require significant amount training data with ground-truth labels to learn model parameters. However, owing scarcity UAV stereo-matching stereo scenarios not fully investigated yet. To facilitate further research, this study proposes pipeline generating accurate dense disparity maps using meshes reconstructed based LiDAR point clouds. Through proposed pipeline, we constructed multi-resolution scenario dataset called UAVStereo, over 34,000 image pairs covering three typical scenes. best our knowledge, UAVStereo first scenarios. The includes synthetic real enable generalization from domain domain. Furthermore, provides multi-scene accommodate various sensors environments. In study, evaluated traditional state-of-the-art methods, highlighting limitations addressing challenges offering suggestions future research. Our available at https://github.com/rebecca0011/UAVStereo.git.