We want evaluation to consist of a set of frames for each keyframes where frame000000.tiff should correspond to the depth map for the Left_Image.png image and frame000001.tiff -> frameN.tiff should correspond to each frame in the video file. The folder structure should match the test set exactly (test_dataset_7/keyframe_1/frame000000.tiff etc).

Just to clarify the format, the submissions should contain the z component of the scene points in mm at each pixel, relative to the left camera. This can be achieved with a single channel float image.

You should submit by October 2nd 23:59 PST by emailing me a dropbox/google drive/alternative link.