DESCRIPTION OF THE CONTENTS

The provided files contain the synthetic database used for training and validation in the paper

    DeepToF: Off-the-Shelf Real-Time Correction of Multipath Interference in Time-of-Flight Imaging
    Julio Marco, Quercus Hernandez, Adolfo Muñoz, Yue Dong, Adrian Jarabo, Min H. Kim, Xin Tong, Diego Gutierrez
    ACM Transactions on Graphics, Vol.36(6)
    
Both dataset files contain a total of 8400 entries in hdf5 format, which provide information of Time-of-Flight
simulations in different diffuse complex scenarios with varying albedos. 

The validation dataset contains 1704 entries. 
The tranining dataset contains 6696 entries.

In particular the datasets contain the following relevant fields (among others) at 256x256 resolution:

/depth:         Time-of-Flight depth (with MPI) obtained by simulating up-to 20-bounce global illumination 
                at different viewpoints of the scenes.
/depth_ref:     Reference depth of the simulated scenes.
/mpi_abs:       Absolute MPI, i.e. depth - depth_ref.
/mpi_rel:       Relative MPI, i.e. (depth - depth_ref) / depth_ref.
/amps:          Amplitude of the simulation as defined by the Time-of-Flight model.

Note that the ToF information in the datasets is generated with perfect sinusoidal modulation frequency at 20MHz.
Any other modulation frequency with different characteristics will affect the shape of the ToF depth and the MPI error.
For more details on the data generation please refer to the full paper

http://webdiis.unizar.es/~juliom/pubs/2017SIGA-DeepToF/

or feel free to contact us by email

Julio Marco
juliom@unizar.es

_______________________________________________________________________________________________________________________
Zaragoza, February 2017