My team and I are proud to announce the release of the new ROad event Awareness for Autonomous Driving (ROAD) Dataset:
https://github.com/gurkirt/road-dataset ROAD is the first benchmark of its kind, designed to allow the community to investigate the use of semantically meaningful representations of dynamic road scenes to facilitate situation awareness and decision making. ROAD is the result of 4 years of joint work by Oxford Brookes University, University of Naples Federico II, ETH Zurich and Mazandaran University and is built upon the Oxford Robotics Institute’s RobotCar dataset. It contains 22 long-duration videos (ca 8 minutes each), ideal for continual learning research, annotated in terms of “road events”, defined as triplets E = (Agent, Action, Location) and represented as ‘tubes’, i.e., a series of frame-wise bounding box detections. ROAD is a large, high-quality multi-label benchmark, with 122K labelled video frames comprising 560K detection bounding boxes associated with 1.7M unique individual labels (560K agent labels, 640K action labels and 499K location labels). ROAD has the ambition to become the reference benchmark for a variety of situation awareness tasks: agent, action and event detection; prediction of intention, trajectories and future events; modelling of complex road activities; instance- and class-incremental continual learning; machine theory of mind capabilities; automated decision making. The GitHub repository contains all the necessary instructions to pre-process the 22 ROAD videos, unpack them to the correct directory structure and run the baseline model, which we termed 3D-RetinaNet and is available at https://github.com/gurkirt/3D-RetinaNet The arXiv report with the full description of dataset, tasks, baseline and result is available here: https://arxiv.org/abs/2102.11585 Please cite the above report when using ROAD in your academic work. For any queries please email Dr Gurkirt Singh ( gurkirt.si...@vision.ee.ethz.ch) or Prof Fabio Cuzzolin ( fabio.cuzzo...@brookes.ac.uk). ---------------------------------------------------- Fabio Cuzzolin Professor of Artificial Intelligence Director of the Visual Artificial Intelligence Laboratory Member of the Board of the Institute for Ethical AI School of Engineering, Computing and Mathematics Oxford Brookes University Oxford, UK http://cms.brookes.ac.uk/staff/FabioCuzzolin/ https://www.linkedin.com/in/fabio-cuzzolin-b481a928/ +44 (0)1865 484526 My office hours are Monday10-11am and Tuesday 10:30-11:30.
_______________________________________________ uai mailing list uai@engr.orst.edu https://it.engineering.oregonstate.edu/mailman/listinfo/uai