Presentation - Computer Science Department, Technion
Download
Report
Transcript Presentation - Computer Science Department, Technion
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Transitive Re-identification
Yulia Brandπ
Tamar Avrahamπ
Michael Lindenbaumπ
πElectrical Engineering Department
πComputer Science Department
Technion - I.I.T. Haifa, Israel
This research was supported by the MAGNET program in the Israeli ministry of industry
and commerce, by the Israeli ministry of science and by the E. and J. Bishop research fund.
BMVC 2013
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
ReIDentification (ReID)
Camera A, time t
Camera B, time t+βt
β’ upper image from: A. Bialkowski, S. Denman, P. Lucey, S. Sridharan, and C. B. Fookes. βA database for person re-identification in multicamera surveillance networks.β (DICTA 2012)
β’ lower images: courtesy of Marco Cristani
Introduction
Motivation
The Transitive Alg.
ReIDentification
Experiments
Summary & Future Work
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
ReIDentification
100
90
ICT β learning based
Recognition %
80
SDALF - direct
70
60
50
40
30
ICT trained on 3-7 data
SDALF for cameras 3-7
20
10
1
2
3
4
5
6
7
rank score
8
9
10
11
12
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
ReIDentification
Camera A
F
same
C
Camera B
F
not same
* Image from: L. Bazzani, M. Cristani, and V. Murino. Symmetry-driven
accumulation of local features for human characterization and reidentification.
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Motivation
πͺπ
πͺπ
πͺπ΅
πͺπ
πͺπ΅
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Motivation
By applying the
transitive algorithm:
πͺπ
By recursively applying the
transitive algorithm:
πͺπ
πͺπ
πͺπ΅
πͺπ
πͺπ΅
πͺπ
πͺπ΅
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ
directly trainable pairs
πͺπ
non-directly trainable pairs
πͺπ
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Motivation
πͺπ
πͺπ΅
πͺπ
πͺπ
πͺπ
πͺπ
πͺπ΅+π
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
πͺπ
πͺπ΅
A
πͺπ
πͺπ
πͺπ
πͺπ
B
πͺπ
C
πͺπ
πͺπ
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
A
πΊπ¨π©
B
πΊπ©πͺ
Training
A
B
C
Test
B
C
A
C
=
?
=
?
=
?
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
A
B
100
90
ICT
Recognition %
80
C
70
SDALF
60
50
Naive ICT
40
30
ICT trained on 3-7 data
SDALF for cameras 3-7
ICT trained on 3-5 and 5-7 data
20
10
1
2
3
4
5
6
7
rank score
8
9
10
11
12
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
π
πΊπ¨π© = {(πππ¨ , ππ© )}
π
ππ¨π© (πππ¨ , ππ© )
π
πΊπ©πͺ = {(πππ© , ππͺ )}
π,
=
A
π,
π=π
π
ππ©πͺ (πππ© , ππͺ )
πβ π
=
π,
π=π
π,
πβ π
πΊπ¨π©
B
πΊπ©πͺ
C
π· ππ¨πͺ ππ¨ , ππͺ =?
π ππ¨πͺ ππ¨ , ππͺ =
π· ππ¨πͺ , ππ¨π© = ππ¨π© , ππ©πͺ = ππ©πͺ , ππ© ππ¨ , ππͺ ) π
ππ©
ππ¨π© β{π,π} ππ©πͺ β{π,π} ππ© βπΉπ
π ππ¨πͺ ππ¨ , ππͺ =
π· ππ¨πͺ , ππ¨π© = ππ¨π© , ππ©πͺ = ππ©πͺ ππ¨ , ππ© , ππͺ )πππ© (ππ© ) π
ππ©
ππ¨π© β{π,π} ππ©πͺ β{π,π} ππ© βπΉπ
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
π
πΊπ¨π© = {(πππ¨ , ππ© )}
π
ππ¨π© (πππ¨ , ππ© )
π,
=
π,
A
π
πΊπ©πͺ = {(πππ© , ππͺ )}
π=π
π
ππ©πͺ (πππ© , ππͺ )
πβ π
=
π,
π=π
π,
πβ π
B
C
π· ππ¨πͺ ππ¨ , ππͺ =?
.
.
.
.
π· ππ¨πͺ ππ¨ , ππͺ =
ππ©
πβ
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ ππΏπ© ππ© π
ππ©
ππ©
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ
ππΏπ© ππ© π
ππ©
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
A
B
π· ππ¨πͺ ππ¨ , ππͺ =
ππ©
πβ
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ ππΏπ© ππ© π
ππ©
ππ©
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ
ππΏπ© ππ© π
ππ©
C
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
The Transitive Algorithm
A
B
π· ππ¨πͺ ππ¨ , ππͺ =
ππ©
πβ
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ ππΏπ© ππ© π
ππ©
ππ©
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ
ππΏπ© ππ© π
ππ©
C
π· ππ¨πͺ ππ¨ , ππͺ β
π
πΊπ©
πβ
π
πΊπ©
ππ© βπΊπ©
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ
ππ© βπΊπ©
π· ππ¨π© ππ¨ , ππ© π· ππ©πͺ ππ© , ππͺ
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Synthetic Experiment 1
ππ© × ππͺ
ππ¨ × ππ©
Recognition
Recognition
%%
ππ¨ × ππͺ
-40 -20
-20
-40
-40 -20
100
100
100
00
0
20 40
40
20
20 40
00
0
-50
-50
-50
00
0
50
50
50
-50
-50
-50
50
50
50
00
000
0
20
20
20
ππ¨ × ππͺ
40
40
40
ππ¨ × ππͺ
60
60
60
-50
-50
-50
00
0
50
50
50
ICT trained
trained on
on AC
AC data
data
ICT
ICT
trained
Naive
ICT on AC data
Naive
ICT
Naive
TRID ICT
TRID
TRID
rank
rank
80
100 rank
80
100
score
score
80
100
score
ππ¨ × ππͺ
π· ππ¨πͺ ππ¨ , ππͺ :
π΅ππππ π°πͺπ»
B
50
50
50
50
50
50
00
0
-50
-50
-50
20
20
20
00
0
-20
-20
-20
A
π°πͺπ» πππππππ
ππ π¨πͺ π
πππ
π»πΉπ°π«
C
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Synthetic Experiment 2
A
ππ© × ππͺ
ππ¨ × π π©
ππ¨ × ππͺ
B
20
20
0
0
-20
-20
Recognition
% %
Recognition
-20
100
0
-20
0
20
50
50
50
0
50
0
0
-50
0
-50
-50
-50
-50
20
0
(d)
0
50
-50
50
C
-50
0
50
-50
0
50
(d)
100
50
50
0
0
0
0
10
20
30
40
50
60
70
10
20
30
40
ππ¨ × ππͺ
50
60
70
ππ¨ × ππͺ
ICT trained on AC data
Naive
ICT on AC data
ICT
trained
TRID ICT
Naive
80
90
100 rank
TRID
score
rank
80
90
100
score
ππ¨ × ππͺ
π· ππ¨πͺ ππ¨ , ππͺ :
π΅ππππ π°πͺπ»
π°πͺπ» πππππππ
ππ π¨πͺ π
πππ
π»πΉπ°π«
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
SAIVT-SoftBio Experiment
β’ image from: A. Bialkowski, S. Denman, P. Lucey, S. Sridharan, and C. B. Fookes. βA database for person re-identification in multi-camera
surveillance networks.β (DICTA 2012)
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
SAIVT-SoftBio Experiment
[A B C] = C3 C5 C7
100
90
90
80
Recognition %
Recognition %
80
70
60
50
40
ICT trained on AC data
Naive ICT
TRID
SDALF
30
20
10
[A B C] = C1 C5 C7
100
0
2
4
6
8
10
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
10
12
0
2
4
6
8
[A B C] = C1 C5 C3
90
90
80
80
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
10
0
2
4
6
8
rank score
12
14
16
18
20
[A B C] = C1 C3 C8
100
Recognition %
Recognition %
100
10
rank score
rank score
10
12
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
14
10
0
5
10
rank score
15
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
SAIVT-SoftBio Experiment
[A B C] = C3 C5 C7
100
90
Recognition %
80
70
60
50
40
ICT trained on AC data
Naive ICT
TRID
SDALF
30
20
10
0
2
4
6
8
10
12
rank score
C
B
A
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
SAIVT-SoftBio Experiment
[A B C] = C3 C5 C7
100
90
90
80
Recognition %
Recognition %
80
70
60
50
40
ICT trained on AC data
Naive ICT
TRID
SDALF
30
20
10
[A B C] = C1 C5 C7
100
0
2
4
6
8
10
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
10
12
0
2
4
6
8
[A B C] = C1 C5 C3
90
90
80
80
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
10
0
2
4
6
8
rank score
12
14
16
18
20
[A B C] = C1 C3 C8
100
Recognition %
Recognition %
100
10
rank score
rank score
10
12
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
14
10
0
5
10
rank score
15
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
SAIVT-SoftBio Experiment
C
B
[A B C] = C1 C3 C8
100
90
80
Recognition %
A
70
60
50
ICT trained on AC data
Naive ICT
TRID
SDALF
40
30
20
10
0
5
10
rank score
15
Introduction
Motivation
Summary
The Transitive Alg.
Experiments
Summary & Future Work
Introduction
Motivation
The Transitive Alg.
Experiments
Summary & Future Work
Future Work
πͺπ
πͺπ
πͺπ
πͺπ
Introduction
Motivation
Thank-You.
The Transitive Alg.
Experiments
Summary & Future Work