Ibims-1, short for Independent benchmark images and matched scans v1, is a new high-quality RGB-D dataset. It is designed for testing single image depth estimation (SIDE) methods. Researchers used a custom acquisition setup consisting of a digital single-lens reflex (DSLR) camera and a high-precision laser scanner to obtain high-resolution images and high-precision depth maps of different indoor scenes. Compared with other RGB-D datasets, the iBims-1 dataset has significant advantages in low noise level, sharp depth transitions, no occlusion, and high depth range.

The dataset contains:

Core dataset: 100 high- and low-resolution RGB-D image pairs of various indoor scenes; masks for invalid, transparent, and planar areas (tables, floors, walls); masks for different depth transforms; camera calibration parameters.

Auxiliary datasets: 56 different color and geometric enhancements of each image of the core dataset; additional hand-held images for testing the MVS method; images of printed patterns and photographs posted on walls to evaluate the performance of textured surfaces; several RGB-D image sequences of static scenes with different illumination.