This page has been created in order to distribute a prototype software
implementing the RGBD-SOBS and RGB-SOBS algorithms presented in [3]:
L. Maddalena, A. Petrosino, Self-Organizing
Background Subtraction Using Color and Depth Data, Multimedia Tools and
Applications, 2018
Click here to download the Windows executable
(WinZip compressed) together with the needed OpenCV .dll's.
Click here to download the Mac executable (Zip
compressed).
If you have problems downloading, please contact lucia.maddalena "at"
cnr.it; if you use the software, please cite the above mentioned paper.
Basic usage for SBM-RGBD dataset:
1)
Unzip the RGBD-SOBS executable file (eventually with the Windows
.dll's) into the directory holding the SBM-RGBD sequences [1].
2)
Place estimated color background images for each sequence (to be adopted
for background initialization as in Eq. (3) of [3])
in the related video directory. You can use those adopted in [3],
which have been built using LabGen [4] over the first
L=100 initial color frames (download them here).
To obtain all masks for the "Bootstrapping/adl24cam0"
sequence, just type:
RGBD-SOBS ./Bootstrapping/adl24cam0
Click here to
download the masks computed by RGBD-SOBS and here to download
the masks computed by RGB-SOBS for the whole dataset.
Usage with generic image sequences:
RGBD-SOBS <SeqName> [Parameters]
where
·
<SeqName>: sequence name (complete path), not including frame
numbers. Image sequences consist of
o
color data saved in .png image files named with consecutive 6
digit numbers in the form “in<number>.png”
(e.g., in000001.png) and stored in
directory <SeqName>/input.
o
depth data saved in .png image files named with consecutive 6 digit
numbers in the form “d<number>.png”
(e.g., d000001.png) and stored in
directory <SeqName>/depth.
o
color background initial estimate stored in directory <SeqName>.
·
[parameters]: optional,
including:
-RGBD #:
To use depth data (1: RGBD-SOBS) or not (0: RGB-SOBS). Default: 1
-nini #:
number of first sequence frame to be considered.
-nend #:
number of last sequence frame to be considered.
-n #: (square root of) number of weight vectors
for each pixel. Default: 3
-K #: Number of initial frames for training.
Default: fromIdx-1 (fromIdx read from file ‘temporalROI.txt’ as in [1])
-e1 #: Distance threshold e1 for training phase (Eq.
(7)). Default: 1.0
-e2 #: Distance threshold e2 for testing phase (Eq.
(7)). Default: 0.008
-c1 #: Learning rate c1 for training phase
(Eq. (11)). Default: 1.0
-c2 #: Learning rate c2 for testing phase
(Eq. (11)). Default: 0.05
-Cw #:
Size of the neighbourhood for Spatial Coherence (Eq. (6)). Default: 5
-s
#: To apply shadow removal (as in [2]).
Default: 1 (apply)
-g #: Shadow detection value for g in Eq. (5) in [2]. Default: 0.7
-b #: Shadow detection value for b in Eq. (5) in [2]. Default: 1.0
-tS #:
Shadow detection value for tS
in Eq. (5) in [2]. Default: 0.1
-tH #: Shadow detection value for tH in Eq. (5) in [2]. Default: 10.0
-ROI #:
To use ROI.bmp mask as in [1]. Default: 1 (do use)
-nameCE
name: To use 'name' as color BG image for initialization (Eq. (3)). Default:
CinitImageL.png (LabGen)
-m #: To save all background model images. Default: 0
(do not save)
-l
#: To save all detection masks (color, depth, fused). Default: 0 (save
only fused)
1) RGBD-SOBS
Provides the above information on usage.
2) RGBD-SOBS c:/Sequences/mysequence
where sequence mysequence has:
- color data saved
in .png image files named in000001.png, …, in00050.png and stored in directory c:/Sequences/mysequence/input.
- depth data saved
in .png image files named d000001.png, …, d00050.png and stored in directory c:/Sequences/mysequence/depth.
- color background
initial estimate MyColorEstimate.png stored in directory
c:/Sequences/mysequence.
This gives the moving
object detection mask for frames from 11 to 50 (named bin000011.png to
bin000011.png) achieved by RGBD-SOBS training the color and background models
on the first 10 frames.
3) RGBD-SOBS c:/Sequences/mysequence
same as before, but
without using depth information (i.e., this is RGB-SOBS).
References:
[1] SBM-RGBD dataset, available
at http://rgbd2017.na.icar.cnr.it/SBM-RGBDdataset.html
[2] L.
Maddalena, A. Petrosino, A Self-Organizing
Approach to Background Subtraction for Visual Surveillance Applications, IEEE Transactions
on Image Processing, DOI 10.1109/TIP.2008.924285, Vol. 17, no. 7, pagg.
1168-1177, July 2008.
[3]
L. Maddalena, A. Petrosino, Self-Organizing Background Subtraction Using Color and Depth Data, Multimedia Tools and Applications, 2018.
[4]
B. Laugraud, S. Pierard, M.
Braham, M. Van Droogenbroeck, Simple
median-based method for stationary background generation using background
subtraction algorithms., In: New Trends in Image Analysis and
Processing-ICIAP 2015 Workshops, LNCS, DOI 10.1007/978-3-319-23222-5_58,
vol. 9281, pp. 477-484. Springer, 2015.
Last update: October
11, 2018.