Facebook Reality Labs has launched the Scene Understanding and Modeling (SUMO) challenge, which targets development of comprehensive 3D scene understanding and modeling algorithms. The challenge has been developed by a team of computer vision researchers at Facebook together with collaborators from Stanford, Princeton and Virginia Tech.

The goal of the SUMO challenge is to generate an instance-based 3D representation of an indoor scene given only a 360-degree RGB-D image taken from a single viewpoint. The generated scene is modeled by a collection of elements, each of which represents one object, such as a wall, the floor, or a chair.

Participants can join in any of the three increasingly detailed and difficult performance tracks: the bounding boxes track, in which the scene is represented by a collection of oriented bounding boxes; the voxels track, where the scene is a collection of oriented voxel grids; and the meshes track, where the scene is a collection of textured surface meshes.

Participants will be evaluated on their ability to consistently infer the correct geometry, pose, appearance and semantics of the elements in each scene.

The challenge will run from August 29th until November 16th, 2018. The top winners in each track will receive prizes, including cash rewards and NVIDIA Titan X GPUs. Winners will be announced at the SUMO Challenge Workshop on December 2nd at ACCV 2018, where they will present their results.

For more information, visit the SUMO Challenge web site.