SUN RGB-D: Results as Published

SUN RGB-D is an RGB-D benchmark for indoor scene understanding. The project page describes about 10,000 images from four sensors, densely labeled with 146,617 2D polygons, 58,657 oriented 3D boxes, a 3D room layout, and a scene category, and the released archive contains 10,335 images. The suite is for training and for evaluation with 3D metrics. The page ships separate evaluators for object detection, holistic scene understanding, and room layout, and it does not name the single number a model card may print.

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.