ToF-RGB Fusion Dataset

This dataset contains depth maps recorded by an ESPROS epc635 Time-of-Flight camera and RGB images from a Raspberry Pi camera module V2, for use in the fusion of these sensors to increase the resolution and framerate of the ToF camera.

The dataset contains three scenes, one of a paper dodecahedron and two of a wooden grid with holes of various sizes, with between 86 and 230 consecutive video frames. In each scene, the camera moves toward the object. Ground truth depth maps were created in Blender and are provided alongside the recorded images. The framerate of the RGB camera is 24 fps. The ToF images are captured at 12 fps but show large fluctuations and are not synchronized with the RGB camera.

The content of the included files are:

ToF Dataset: Recorded HDR ToF camera depth maps (160x60), ground truth depth maps (2560x960), and recorded RGB images (2560x960). The depth maps contain the depth values in millimeters, stored as 16-bit PNG files. RGB and ground truth depth maps are named continuously by their frame number. The ToF depth maps are named with a frame number, based on the closest RGB image in the time domain.

Additional Data: All data used in the creation of the ToF dataset (All recorded depth maps and color images, calibration images of a 4x5 checkerboard calibration pattern with a square size of 43mm, calibration parameters, Blender files)