Each zip file shared on the download webpage corresponds to a Sequence, i.e., a set of synchronized videos. Each video corresponds to the information of one camera. These videos are encoded with H.265 and has the following format:
The left half of the image contains the information of, what we call in the following, the front fisheye lens, whereas the right half describes the image captured by the so-called rear fisheye lens. All the geometrical deformations of the fish eye capture have been estimated (according to the unified spherical model) and are explained here. We precise that we have chosen not to share the stitched images because they lose the geometrical model that have been learned during the calibration. However, they could be easily retrieved from the unified spherical model and its estimated parameters.
