The SUMO Challenge

We introduce the Scene Understanding and Modeling (SUMO) challenge with the goal of studying and evaluating the performance of RGBD-to-3D semantic scene modeling algorithms. Challenge participants are tasked to derive a complete, instance-based 3D representation of a scene given an RGB-D representation. The complete representation should highlight geometric instances, appearance, and semantics. This document describes the details of the challenge tasks, including scene elements of interest, data format, and metrics.