3D BAT: A Semi-Automatic, Web-based 3D Annotation Toolbox for Full-Surround, Multi-Modal Data Streams

In this paper, we focus on obtaining 2D and 3D labels, as well as track IDs\nfor objects on the road with the help of a novel 3D Bounding Box Annotation\nToolbox (3D BAT). Our open source, web-based 3D BAT incorporates several smart\nfeatures to improve usability and efficiency. For instance, this annotation\ntoolbox supports semi-automatic labeling of tracks using interpolation, which\nis vital for downstream tasks like tracking, motion planning and motion\nprediction. Moreover, annotations for all camera images are automatically\nobtained by projecting annotations from 3D space into the image domain. In\naddition to the raw image and point cloud feeds, a Masterview consisting of the\ntop view (bird's-eye-view), side view and front views is made available to\nobserve objects of interest from different perspectives. Comparisons of our\nmethod with other publicly available annotation tools reveal that 3D\nannotations can be obtained faster and more efficiently by using our toolbox.\n