LBVS-HDR: Learning Based Visual Saliency prediction for High Dynamic Range Video
LBVS-HDR: is
a Visual Attention Model (VAM) for High Dynamic Range (HDR) video data. The
original paper that introduces the LBVS-HDR is [1]. Reference [2] is on the usage
of LBVS-HDR resulting saliency maps for quality assessment of HDR video. Any
work that uses these data is required to cite the following related papers:
[1] A. Banitalebi-Dehkordi, Y.
Dong, M. T. Pourazad, and Panos
Nasiopoulos, “A Learning Based Visual Saliency
Fusion Model For High Dynamic Range Video (LBVS-HDR),” 23rd European
Signal Processing Conference, EUSIPCO 2015.
[2] A. Banitalebi-Dehkordi1, M. Azimi,
M. T. Pourazad, and P. Nasiopoulos1,
"Saliency-Aided HDR Quality Metrics," ISO/IEC JTC1/SC29/WG11
MPEG2014/m37317, Oct. 2015, Geneva, Switzerland.
HDR video dataset: The four video sequences used
in this dataset are: “Balloons”, “Fire_eater”,
“Market”, and “Tibul”, which
are all being used for MPEG HDR video compression standardization activities.
Refer to [1] and [2] for more information about the videos. The source files of
the videos are licensed to elsewhere. However, we have provided a link for each
sequence to obtain the data from there.
Saliency maps:
The following saliency maps are generated using the LBVS-HDR
for the mentioned videos.
Disclaimer: This dataset is free for use only for research
purposes. For any commercial use please contact us at: “panos-at-ece.ubc.ca”,
or “dehkordi-at-ece.ubc.ca”
HDR saliency maps for the “Balloons” sequence: 125 MB
Source: CableLabs
Also introduced here: (Daryl Malas, Ren´e Van der Bleuten, Edouard Francois,
Koohyar Minoo, and Elena Alshina, “Cablelabs content
input hdr wcg,” in ISO/IEC JTC1/SC29/WG11
MPEG2014/ m35465, Lausanne, CH, 2014)
HDR saliency maps for the “Tibul”
sequence: 155 MB Source: Technicolor (Y. Olivier, D. Touzé, C. Serre, S. Lasserre, F. L. Léannec and E. François, “Description
of HDR sequences,” ISO/IEC JTC1/SC29/WG11 MPEG2014/m31957, Oct 2014.) HDR saliency maps for the “Fire Eater” sequence:
100 MB Source: Technicolor (Y. Olivier, D. Touzé, C. Serre, S. Lasserre, F. L. Léannec and E. François, “Description
of HDR sequences,” ISO/IEC JTC1/SC29/WG11 MPEG2014/m31957, Oct 2014.) HDR saliency maps for the “Market” sequence: 255 MB Source: Technicolor (Y. Olivier, D. Touzé, C. Serre, S. Lasserre, F. L. Léannec and E. François, “Description
of HDR sequences,” ISO/IEC JTC1/SC29/WG11 MPEG2014/m31957, Oct 2014.) Eye-Tracking data: DML-iTrack-HDR Eye-tracking data that is used for training the LBVS-HDR is
explained here, and can be download
from the links below: Fixation Density
Maps for different sequences:
