Robust Visual Tracking – Algorithms, Evaluations and Problems

Robust Visual Tracking – Algorithms, Evaluations and

Problems

Haibin LingDepartment of Computer and Information Sciences

Temple University

Philadelphia, PA 19122

October 15, 2014

Visual Tracking

Single target tracking (model-free) (PAMI’11,CVPR’11,ICCV’11,CVPR’12,ICCV’13,ECCV’14)

Pose tracking (Sigal et al 2004)

Contour tracking (CVPR’14b)

Continuously localization of a visual entity or visual entities.

Multi-target tracking (CVPR’13,CVPR’14a)

Visual TrackingContinuously localization of a visual entity or visual entities.

Related work- Tooooooo many to be listed- A survey by Yilmaz, Javed & Shah in 2006- There are many influential trackers after 2006

Single target tracking (model-free) (PAMI’11,CVPR’11,ICCV’11,CVPR’12,ICCV’13,ECCV’14)

Outline

• Problem formulation and particle filter tracking framework

• Visual tracking using sparse representation

• Reducing bias in tracking evaluation

• Recent and future work

Problem formulation

Input:

• A sequence of images: I0, I1, …, It, …

• Target of interest at the initial frame: x0

A target is represented by a state vector

x = (pos, scale, orientation)‘

Output:• Targets in each of the following frames

– x1, …, xt, …

Tracking by Bayesian Estimation

At frame t, find the best xt by Bayesian inference

Using observations (features) extracted from images I0, I1, …, It :

We have

Kalman filter– Gaussian everywhere closed form solution – But, probabilities in visual tracking is not usually Gaussian

Particle filter– Probability propagation: iterative prediction and updating – Sampling techniques

),,...,,|(maxarg 011 IIIIxpx tttx

tt

)|(maxarg :0 ttx

t yxpxt

},...,,{:;,...,, 10:010 ttt yyyyyyy

Bayesian estimation:

Particle Filter (Isard & Blake 98)

Prediction:

Update:

)|(maxarg :0 ttx

t yxpxt

Visual tracking

1

11:0111:0 )|()|()|(tx ttttttt dxyxpxxpyxp

likelihood nobservatio:)|(

etc) motion, (drift,y probabilit n transitiostate:)|( 1

tt

tt

xyp

xxp

)|()|()|( 1:0:0 tttttt yxpxypyxp

Probability propagation

Particle sampling (sequential Monte Carlo)

Approximate the posterior density by a set of weighted samples:

Niwx it

it ,...,2,1:),( )()(

).|( e.g.,, particlefor weight theiswhere it

it

it

it xypxw

Now we need to decide

Outline

• Problem formulation and particle filter tracking framework

• Visual tracking using sparse representation

• Reducing bias in tracking evaluation

• Recent and future work

MotivationIntuition• During tracking, there is a large redundancy in the observation of target

appearance• It is common to represent the target appearance using a linear

representation

Idea• Introduce sparse constraints in the linear target representation• Non-negativity constraints

Advantage• Models observation redundancy naturally.• Addresses discrete appearance corruption such as occlusion (Wright et al.

2009) • Benefits from recent advance in solutions for sparse coding/compressive

sensing (Candes et al. 2006, Donoho 2006)• A flexible framework (as illustrated in many extensions)

Sparse Representation for Tracking

• A candidate y approximately lies in a linear subspace, which is spanned by templates from past observation

e

a]I,T[ddnn eeeaaa iiittty 22112211

nnaaa ttty 2211 nnaaa ttty 2211

Task: find a sparse solution for a and e.

Rewrite as

Non-negativity Constraints

• In addition to the (positive) trivial templates I, we include negative trivial templates -I.

0cBc,ˆ

e

e

a

]II,,T[y

-

)i()i()i( d2211 deee

ddnn eeeaaa iiittty 22112211

where ai, ei, ei- >=0 .

The formula can be rewritten as

Example Templates

y

e

e

a

B c

Comparing Good and Bad Candidates

Achieving Sparse Solutions

0c,cyBcmin0

2

2

Our task is to find a sparse solution to the following linear system,

0cBc,y

It leads to an L0 minimization task, such as

This can be well approximated, under very flexible conditions, by an L1 minimization,

0c,cyBcmin1

2

2

Extension• Speed up

– Speed up: bounded particle resampling (CVPR’11)– Speed up: accelerated proximal gradient (CVPR’12)– Blurred target tracking (ICCV’11)

• Other sparse-representation trackers– Liu et al. ECCV'10, – Li, Shen & Shi CVPR'11, Liu et al CVPR'11, Kwak et al ICCV’11– Zhong, Lu & Yang CVPR'12; Jia, Lu & Yang CVPR'12; Zhang,

Zhang & Yang CVPR'12; ZhangT et al CVPR'12, – ZhangT et al IJCV’13, Hu et al PAMI’14– …

Outline

• Problem formulation and particle filter tracking framework

• Visual tracking using sparse representation

• Reducing bias in tracking evaluation

• Recent and future work

Reducing Subjective Bias

• Which are the best trackers among all?• Implementing and testing on a large benchmark (e.g.,

Wu et al 2013) is a huge project.

• Recent trend: compare the authors’ own tracker with many other trackers.

• Their own tracker typically performs the best.– It has advantages that the authors want to highlight.– Optimizing all trackers is non-trivial, if not possible.

• We aim to reduce such biases and provide a more practical comparison.

An example

A B C D E

Seq 1 17.5 56.7 11.3 10.5 5.0

Seq 2 7.0 39.2 8.5 39.2 6.1

… … … … … …

Seq N 30.7 66.2 20.4 120.4 24.9

• The best two results are shown in red and blue

Average Center Location Error

The proposed tracker

The authors’ previous tracker

Partial ranking representation

A B C D E

Seq 1 17.5 56.7 11.3 10.5 5.0

Seq 2 7.0 39.2 8.5 39.2 6.1

… … … … … …

Seq N 30.7 66.2 20.4 120.4 24.9

Average Center Location Error

Higher rank Lower rank

D < A < B

Higher rank Lower rank

D < A < B

A < B = D

… < … < …

A < B < D

D10.5

A17.5

B56.7

< <

A B C D E

Seq 1 17.5 56.7 11.3 10.5 5.0

Seq 2 7.0 39.2 8.5 39.2 6.1

… … … … … …

Seq N 30.7 66.2 20.4 120.4 24.9

Average Center Location Error

Pairwise representation

(A, B, 1)

(D, A, 1)

(D, B, 1)

(A, B, 1)

(A, D, 1)

(B, D, 0.5)

Seq 1 Seq 2

(D, B, 0.5)

…

(A, B, 1)

(A, D, 1)

(B, D, 1)

Seq N

A7.0

B39.2

<

D39.2

=

Data Statistics

• PAMI (2000 Vol.22– 2013 Vol.35),

IJCV (2000 Vol.36 – 2013 Vol.104)• ICCV, CVPR, ECCV (2005 – 2013)• 45 papers (tournament) contain useful table data• 48 trackers appear in the data at the first stage• 15 trackers are left after the cleaning• 664 partial rankings• 6280 pairs of records with 151 draw records

Paper selection and data cleaning

• More than 2 trackers left after remove unqualified trackers

• Independent assumption– Conference to journal extension– Duplicate experimental results

• Significant lack of data– Compared only in one tournament– #records ≤ 10

Rank aggregation• Rank aggregation (Ailon 2010)

– Find a full-ranking to minimize the total violation of pairwise comparison.

– NP-Hard, LpKwikSorth algorithm

• PageRank-like ranking (Page et al. 1999)– Graph-based solution

• Elo’s rating (Elo 1978)– Widely used in sport ranking (chess, football, …)– Sequentially update score based on each game

• Glicko’s rating (Glickman 1999)– Extension of Elo’s rating by introducing confidence

Ranking results

Outline

• Problem formulation and particle filter tracking framework

• Visual tracking using sparse representation

• Reducing bias in tracking evaluation

• Recent and future work

Tracking with GPR (TGPR)Transfer Learning Based Visual Tracking with Gaussian Processes Regression

Gao, Ling, Hu & Xing, ECCV 2014

Source code of TGPR available: http://www.dabi.temple.edu/~hbling/code/TGPR.htm or http://jingao.weebly.com/

Promising ResultsCVPR2013 Benchmark

(Wu et al 2013)

50 sequences

Princeton Benchmark (Song & Xiao 2013)

100 sequences

VOT2013 (Kristan et al

2013)

16 sequences

Acknowledgement

• CollaboratorsChenglong Bao, Erik Blasch, Jin Gao, Weiming Hu

Hui Ji, Xue Mei, Yu Pang, Yi Wu

• Funding• National Sciences Foundation

• Air Force Research Laboratory

Thank You!&

Questions?