SlideShare a Scribd company logo
Multi-Camera Multi-Human
Tracking System
Copyright © Yu-Sheng Chen [Yosen]
Outline
Top view of the entire tracking system
What local trackers do
Identify people depth order
Handle body occlusion
What central server does
Compare observations across cameras
Classify to different people
Recognize people when re-entering
Yosen Chen, Copyright Reserved 2
Images
Local camera
#A tracker
 ,
1
A
tz 2
A
tz
2
B
tz
1
B
tz
Local camera
#B tracker
Images
Central Server
multi-camera correspondence
Person#1 Person#2
Identify how many people
in the observed space
Shared data on Inter-process platform
 ,2
A
tz 1
B
tz 1
A
tz 2
B
tz
Listen/
Broadcast
What local trackers do?
Multi-human
tracking &
detection
Central
Server
Operation flow of local trackers
Identify
people depth
order
If detect new,
broadcast
Get people labels
& people heights.
Handle body
occlusion
Start
Keep updating body info. to server databases
Extract
people body
info.
Local tracker
Central server
What local trackers do:
 Identify depth order
A. By appearances
B. By standing locations
Image source
1:Front
2:Rear
Depth order estimation by head shape
completeness is a bad idea!
Yosen Chen, Copyright Reserved 6
What local trackers do:
 Identify depth order
A. By appearances
B. By standing locations
Image source
1:Front
2:Rear
Yosen Chen, Copyright Reserved 8
Tell the depth order by people’s
standing locations
How to know the standing locations?
 To be discussed in the central server part!!
Bd Ground plane
Ad
Camera image
Cd
BdAd Cd depth order: A=1, B=2, C=3
A
B
C
Depth order estimation by 3D standing
locations is accurate.
Yosen Chen, Copyright Reserved 9
What local trackers do:
 Handle body occlusion
 Occluded parts can be estimated by
3D geometry with body modeling
Occlusion Estimation by 3D Body
Modeling
Yosen Chen, Copyright Reserved 11
Ground plane
Camera image
B
A
AB
Camera visible parts
Camera invisible parts
Body Occlusion Handling
Yosen Chen, Copyright Reserved 12
Depth order
check
Body occlusion
handling
Target info
storage
Body appearance is available for
color extraction
Body appearance is not available
What central server does?
Compare
observations
across cameras
Local
trackers
Operation flow of central server
Get observation
correspondence
between cameras
If detect new,
broadcast
Return people labels & people heights.
Start
Keep updating body info. to server database
People
database
Compare with
database to recognize
/create people
Local tracker
Central server
What central server does:
 Compare observations across
cameras
 Geometric-based color comparison
 2D position on image plane  3D line in real space
 3D standing point in real space  2D feet position on image plane
observation r
Optic center Image plane of
camera j
Ground plane
Optic center
observation q
Image plane of
camera k
3D Geometric Correspondence across
Cameras
Calibrated camera system
Head Point
height
Estimated Standing Location
Yosen Chen, Copyright Reserved 15
What central server does:
 Classify to different people
 Use decisions in higher confidence to
determine lower ones.
Observation Correspondence
across Cameras
Object Conf = 3
Object Conf = 2
Matching body color
(Block-based body color comparison)
Yosen Chen, Copyright Reserved 17
Different colors mean different labels
Multi-People Correspondence
Yosen Chen, Copyright Reserved 18
1. Multi-people depth order √
2. body occlusion handling √
3. Multi-people correspondence √
What central server does:
 Identify people if re-enter
 Construct people database once
they got tracked.
Yosen Chen, Copyright Reserved 20
Check height
difference
How to recognize people when they re-enter
Are they all from the same person?
Body Color
comparison
by block
Label history
consistency
No, if diff > threshold No, if color unmatched
Yes, they are from
the same person!!
Camera
observations
People model
in database
Future Works…
Extend to more cameras
Challenge of
Communication load & algorithm complexity
Tracking/labeling stability
More challenges:
Outdoors Surveillance (human body detection)
Uncalibrated multi-camera system (Training)
Motive cameras (rotation 3D-2D mapping)
Yosen Chen, Copyright Reserved 21
Thanks for
Pf. Li-Chen Fu
Advanced Control Lab, NTU
Intelligent Robot Lab, NTU
Copyright © Yu-Sheng Chen [Yosen]

More Related Content

Viewers also liked

People or human tracking system
People or human tracking systemPeople or human tracking system
People or human tracking system
Venkatesan S
 
Video Inpainting detection using inconsistencies in optical Flow
Video Inpainting detection using inconsistencies in optical FlowVideo Inpainting detection using inconsistencies in optical Flow
Video Inpainting detection using inconsistencies in optical Flow
Cybersecurity Education and Research Centre
 
Seminar On Kalman Filter And Its Applications
Seminar On  Kalman  Filter And Its ApplicationsSeminar On  Kalman  Filter And Its Applications
Seminar On Kalman Filter And Its Applications
Barnali Dey
 
Kalman filters
Kalman filtersKalman filters
Kalman filters
Saravanan Natarajan
 
Kalman filter
Kalman filterKalman filter
Kalman filter
Raghava Raghu
 
Kalman filter for object tracking
Kalman filter for object trackingKalman filter for object tracking
Kalman filter for object tracking
Mohit Yadav
 
Kalman filter - Applications in Image processing
Kalman filter - Applications in Image processingKalman filter - Applications in Image processing
Kalman filter - Applications in Image processing
Ravi Teja
 
Image segmentation
Image segmentationImage segmentation
Image segmentation
Rania H
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
Raviraj singh shekhawat
 
Background subtraction
Background subtractionBackground subtraction
Background subtraction
Shashank Dhariwal
 
Object Recognition
Object RecognitionObject Recognition
Object Recognition
Eman Abed AlWahhab
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
Manav Mittal
 
IMAGE SEGMENTATION.
IMAGE SEGMENTATION.IMAGE SEGMENTATION.
IMAGE SEGMENTATION.
Tawose Olamide Timothy
 
Image segmentation ppt
Image segmentation pptImage segmentation ppt
Image segmentation ppt
Gichelle Amon
 
Human Re-identification using Soft Biometrics in Video Surveillance
Human Re-identification using Soft Biometrics in Video SurveillanceHuman Re-identification using Soft Biometrics in Video Surveillance
Human Re-identification using Soft Biometrics in Video Surveillance
Shengzhe Li
 
Background subtraction
Background subtractionBackground subtraction
Background subtraction
Raviraj singh shekhawat
 

Viewers also liked (16)

People or human tracking system
People or human tracking systemPeople or human tracking system
People or human tracking system
 
Video Inpainting detection using inconsistencies in optical Flow
Video Inpainting detection using inconsistencies in optical FlowVideo Inpainting detection using inconsistencies in optical Flow
Video Inpainting detection using inconsistencies in optical Flow
 
Seminar On Kalman Filter And Its Applications
Seminar On  Kalman  Filter And Its ApplicationsSeminar On  Kalman  Filter And Its Applications
Seminar On Kalman Filter And Its Applications
 
Kalman filters
Kalman filtersKalman filters
Kalman filters
 
Kalman filter
Kalman filterKalman filter
Kalman filter
 
Kalman filter for object tracking
Kalman filter for object trackingKalman filter for object tracking
Kalman filter for object tracking
 
Kalman filter - Applications in Image processing
Kalman filter - Applications in Image processingKalman filter - Applications in Image processing
Kalman filter - Applications in Image processing
 
Image segmentation
Image segmentationImage segmentation
Image segmentation
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
 
Background subtraction
Background subtractionBackground subtraction
Background subtraction
 
Object Recognition
Object RecognitionObject Recognition
Object Recognition
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
 
IMAGE SEGMENTATION.
IMAGE SEGMENTATION.IMAGE SEGMENTATION.
IMAGE SEGMENTATION.
 
Image segmentation ppt
Image segmentation pptImage segmentation ppt
Image segmentation ppt
 
Human Re-identification using Soft Biometrics in Video Surveillance
Human Re-identification using Soft Biometrics in Video SurveillanceHuman Re-identification using Soft Biometrics in Video Surveillance
Human Re-identification using Soft Biometrics in Video Surveillance
 
Background subtraction
Background subtractionBackground subtraction
Background subtraction
 

Similar to Multi-Camera Multi-Human Tracking System (oral presentation)

Presentation of Visual Tracking
Presentation of Visual TrackingPresentation of Visual Tracking
Presentation of Visual Tracking
Yu-Sheng (Yosen) Chen
 
A Comparison of People Counting Techniques via Video Scene Analysis
A Comparison of People Counting Techniques viaVideo Scene AnalysisA Comparison of People Counting Techniques viaVideo Scene Analysis
A Comparison of People Counting Techniques via Video Scene Analysis
Poo Kuan Hoong
 
Wang midterm-defence
Wang midterm-defenceWang midterm-defence
Wang midterm-defence
Zhipeng Wang
 
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
patrickrobertson
 
When Remote Sensing Meets Artificial Intelligence
When Remote Sensing Meets Artificial IntelligenceWhen Remote Sensing Meets Artificial Intelligence
When Remote Sensing Meets Artificial Intelligence
WahyuRahmaniar2
 
Motivation for image fusion
Motivation for image fusionMotivation for image fusion
Motivation for image fusion
VIVEKANAND BONAL
 
Basic principle of C.T.
Basic principle of C.T.Basic principle of C.T.
Basic principle of C.T.
BISHAL KHANAL
 
4. THREE DIMENSIONAL DISPLAY METHODS
4.	THREE DIMENSIONAL DISPLAY METHODS4.	THREE DIMENSIONAL DISPLAY METHODS
4. THREE DIMENSIONAL DISPLAY METHODS
SanthiNivas
 

Similar to Multi-Camera Multi-Human Tracking System (oral presentation) (8)

Presentation of Visual Tracking
Presentation of Visual TrackingPresentation of Visual Tracking
Presentation of Visual Tracking
 
A Comparison of People Counting Techniques via Video Scene Analysis
A Comparison of People Counting Techniques viaVideo Scene AnalysisA Comparison of People Counting Techniques viaVideo Scene Analysis
A Comparison of People Counting Techniques via Video Scene Analysis
 
Wang midterm-defence
Wang midterm-defenceWang midterm-defence
Wang midterm-defence
 
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
Improving Simultaneous Localization and Mapping for Pedestrian Navigation and...
 
When Remote Sensing Meets Artificial Intelligence
When Remote Sensing Meets Artificial IntelligenceWhen Remote Sensing Meets Artificial Intelligence
When Remote Sensing Meets Artificial Intelligence
 
Motivation for image fusion
Motivation for image fusionMotivation for image fusion
Motivation for image fusion
 
Basic principle of C.T.
Basic principle of C.T.Basic principle of C.T.
Basic principle of C.T.
 
4. THREE DIMENSIONAL DISPLAY METHODS
4.	THREE DIMENSIONAL DISPLAY METHODS4.	THREE DIMENSIONAL DISPLAY METHODS
4. THREE DIMENSIONAL DISPLAY METHODS
 

Recently uploaded

Barbie Made To Move Skin Tones Matches.pptx
Barbie Made To Move Skin Tones Matches.pptxBarbie Made To Move Skin Tones Matches.pptx
Barbie Made To Move Skin Tones Matches.pptx
LinaCosta15
 
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvnFinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
abbieharman
 
FinalLessonPlanResponding.docxnknknknknknk
FinalLessonPlanResponding.docxnknknknknknkFinalLessonPlanResponding.docxnknknknknknk
FinalLessonPlanResponding.docxnknknknknknk
abbieharman
 
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
taqyea
 
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
19rmjonz
 
Codes n Conventions Website Media studies.pptx
Codes n Conventions Website Media studies.pptxCodes n Conventions Website Media studies.pptx
Codes n Conventions Website Media studies.pptx
ZackSpencer3
 
一比一原版(BC毕业证)波士顿学院毕业证如何办理
一比一原版(BC毕业证)波士顿学院毕业证如何办理一比一原版(BC毕业证)波士顿学院毕业证如何办理
一比一原版(BC毕业证)波士顿学院毕业证如何办理
40fortunate
 
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
Cottage9 Enterprises
 
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
tc73868
 
Minor_Michael_PowerPoint-Presentation.pptx
Minor_Michael_PowerPoint-Presentation.pptxMinor_Michael_PowerPoint-Presentation.pptx
Minor_Michael_PowerPoint-Presentation.pptx
MichaelMinor19
 
Cherries 32 collection of colorful paintings
Cherries 32 collection of colorful paintingsCherries 32 collection of colorful paintings
Cherries 32 collection of colorful paintings
sandamichaela *
 
ARNAUVALERY RECORD STORE SCAVENGER HUNT.
ARNAUVALERY RECORD STORE SCAVENGER HUNT.ARNAUVALERY RECORD STORE SCAVENGER HUNT.
ARNAUVALERY RECORD STORE SCAVENGER HUNT.
ValeryArnau
 
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
➒➌➎➏➑➐➋➑➐➐Dpboss Matka Guessing Satta Matka Kalyan Chart Indian Matka
 
❼❷⓿❺❻❷❽❷❼❽ Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
❼❷⓿❺❻❷❽❷❼❽  Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...❼❷⓿❺❻❷❽❷❼❽  Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
❼❷⓿❺❻❷❽❷❼❽ Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
 
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
 
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
homgo
 
Tibbetts_HappyAwesome_NewArc Sketch to AI
Tibbetts_HappyAwesome_NewArc Sketch to AITibbetts_HappyAwesome_NewArc Sketch to AI
Tibbetts_HappyAwesome_NewArc Sketch to AI
Todd Tibbetts
 
My storyboard for a sword fight scene with lightsabers
My storyboard for a sword fight scene with lightsabersMy storyboard for a sword fight scene with lightsabers
My storyboard for a sword fight scene with lightsabers
AlejandroGuarnGutirr
 
FinalFinalSelf-PortraiturePowerPoint.pptx
FinalFinalSelf-PortraiturePowerPoint.pptxFinalFinalSelf-PortraiturePowerPoint.pptx
FinalFinalSelf-PortraiturePowerPoint.pptx
abbieharman
 
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
a0pr7yf1
 

Recently uploaded (20)

Barbie Made To Move Skin Tones Matches.pptx
Barbie Made To Move Skin Tones Matches.pptxBarbie Made To Move Skin Tones Matches.pptx
Barbie Made To Move Skin Tones Matches.pptx
 
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvnFinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
FinalA1LessonPlanMaking.docxdvdnlskdnvsldkvnsdkvn
 
FinalLessonPlanResponding.docxnknknknknknk
FinalLessonPlanResponding.docxnknknknknknkFinalLessonPlanResponding.docxnknknknknknk
FinalLessonPlanResponding.docxnknknknknknk
 
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
一比一原版加拿大多伦多大学毕业证(uoft毕业证书)如何办理
 
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
哪里办理(sjsu毕业证书)美国圣何塞州立大学毕业证硕士文凭证书原版一模一样
 
Codes n Conventions Website Media studies.pptx
Codes n Conventions Website Media studies.pptxCodes n Conventions Website Media studies.pptx
Codes n Conventions Website Media studies.pptx
 
一比一原版(BC毕业证)波士顿学院毕业证如何办理
一比一原版(BC毕业证)波士顿学院毕业证如何办理一比一原版(BC毕业证)波士顿学院毕业证如何办理
一比一原版(BC毕业证)波士顿学院毕业证如何办理
 
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
Tanjore Painting: Rich Heritage and Intricate Craftsmanship | Cottage9
 
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
哪里购买美国乔治城大学毕业证硕士学位证书原版一模一样
 
Minor_Michael_PowerPoint-Presentation.pptx
Minor_Michael_PowerPoint-Presentation.pptxMinor_Michael_PowerPoint-Presentation.pptx
Minor_Michael_PowerPoint-Presentation.pptx
 
Cherries 32 collection of colorful paintings
Cherries 32 collection of colorful paintingsCherries 32 collection of colorful paintings
Cherries 32 collection of colorful paintings
 
ARNAUVALERY RECORD STORE SCAVENGER HUNT.
ARNAUVALERY RECORD STORE SCAVENGER HUNT.ARNAUVALERY RECORD STORE SCAVENGER HUNT.
ARNAUVALERY RECORD STORE SCAVENGER HUNT.
 
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
Dpboss Matka Guessing Satta Matta Matka Kalyan panel Chart Indian Matka Dpbos...
 
❼❷⓿❺❻❷❽❷❼❽ Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
❼❷⓿❺❻❷❽❷❼❽  Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...❼❷⓿❺❻❷❽❷❼❽  Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
❼❷⓿❺❻❷❽❷❼❽ Dpboss Matka ! Fix Satta Matka ! Matka Result ! Matka Guessing ! ...
 
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
❼❷⓿❺❻❷❽❷❼❽ Dpboss Kalyan Satta Matka Guessing Matka Result Main Bazar chart
 
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
一比一原版美国亚利桑那大学毕业证(ua毕业证书)如何办理
 
Tibbetts_HappyAwesome_NewArc Sketch to AI
Tibbetts_HappyAwesome_NewArc Sketch to AITibbetts_HappyAwesome_NewArc Sketch to AI
Tibbetts_HappyAwesome_NewArc Sketch to AI
 
My storyboard for a sword fight scene with lightsabers
My storyboard for a sword fight scene with lightsabersMy storyboard for a sword fight scene with lightsabers
My storyboard for a sword fight scene with lightsabers
 
FinalFinalSelf-PortraiturePowerPoint.pptx
FinalFinalSelf-PortraiturePowerPoint.pptxFinalFinalSelf-PortraiturePowerPoint.pptx
FinalFinalSelf-PortraiturePowerPoint.pptx
 
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
哪里购买(ucr毕业证书)美国加州大学河滨分校毕业证研究生文凭证书原版一模一样
 

Multi-Camera Multi-Human Tracking System (oral presentation)

  • 2. Outline Top view of the entire tracking system What local trackers do Identify people depth order Handle body occlusion What central server does Compare observations across cameras Classify to different people Recognize people when re-entering Yosen Chen, Copyright Reserved 2
  • 3. Images Local camera #A tracker  , 1 A tz 2 A tz 2 B tz 1 B tz Local camera #B tracker Images Central Server multi-camera correspondence Person#1 Person#2 Identify how many people in the observed space Shared data on Inter-process platform  ,2 A tz 1 B tz 1 A tz 2 B tz Listen/ Broadcast
  • 4. What local trackers do? Multi-human tracking & detection Central Server Operation flow of local trackers Identify people depth order If detect new, broadcast Get people labels & people heights. Handle body occlusion Start Keep updating body info. to server databases Extract people body info. Local tracker Central server
  • 5. What local trackers do:  Identify depth order A. By appearances B. By standing locations Image source 1:Front 2:Rear
  • 6. Depth order estimation by head shape completeness is a bad idea! Yosen Chen, Copyright Reserved 6
  • 7. What local trackers do:  Identify depth order A. By appearances B. By standing locations Image source 1:Front 2:Rear
  • 8. Yosen Chen, Copyright Reserved 8 Tell the depth order by people’s standing locations How to know the standing locations?  To be discussed in the central server part!! Bd Ground plane Ad Camera image Cd BdAd Cd depth order: A=1, B=2, C=3 A B C
  • 9. Depth order estimation by 3D standing locations is accurate. Yosen Chen, Copyright Reserved 9
  • 10. What local trackers do:  Handle body occlusion  Occluded parts can be estimated by 3D geometry with body modeling
  • 11. Occlusion Estimation by 3D Body Modeling Yosen Chen, Copyright Reserved 11 Ground plane Camera image B A AB Camera visible parts Camera invisible parts
  • 12. Body Occlusion Handling Yosen Chen, Copyright Reserved 12 Depth order check Body occlusion handling Target info storage Body appearance is available for color extraction Body appearance is not available
  • 13. What central server does? Compare observations across cameras Local trackers Operation flow of central server Get observation correspondence between cameras If detect new, broadcast Return people labels & people heights. Start Keep updating body info. to server database People database Compare with database to recognize /create people Local tracker Central server
  • 14. What central server does:  Compare observations across cameras  Geometric-based color comparison
  • 15.  2D position on image plane  3D line in real space  3D standing point in real space  2D feet position on image plane observation r Optic center Image plane of camera j Ground plane Optic center observation q Image plane of camera k 3D Geometric Correspondence across Cameras Calibrated camera system Head Point height Estimated Standing Location Yosen Chen, Copyright Reserved 15
  • 16. What central server does:  Classify to different people  Use decisions in higher confidence to determine lower ones.
  • 17. Observation Correspondence across Cameras Object Conf = 3 Object Conf = 2 Matching body color (Block-based body color comparison) Yosen Chen, Copyright Reserved 17 Different colors mean different labels
  • 18. Multi-People Correspondence Yosen Chen, Copyright Reserved 18 1. Multi-people depth order √ 2. body occlusion handling √ 3. Multi-people correspondence √
  • 19. What central server does:  Identify people if re-enter  Construct people database once they got tracked.
  • 20. Yosen Chen, Copyright Reserved 20 Check height difference How to recognize people when they re-enter Are they all from the same person? Body Color comparison by block Label history consistency No, if diff > threshold No, if color unmatched Yes, they are from the same person!! Camera observations People model in database
  • 21. Future Works… Extend to more cameras Challenge of Communication load & algorithm complexity Tracking/labeling stability More challenges: Outdoors Surveillance (human body detection) Uncalibrated multi-camera system (Training) Motive cameras (rotation 3D-2D mapping) Yosen Chen, Copyright Reserved 21
  • 22. Thanks for Pf. Li-Chen Fu Advanced Control Lab, NTU Intelligent Robot Lab, NTU
  • 23. Copyright © Yu-Sheng Chen [Yosen]