advances in face recognition...the face recognition company advances in face recognition research...
TRANSCRIPT
![Page 1: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/1.jpg)
The face recognition company
Advances in Face Recognition Research
Presentation for the 2nd End User Group MeetingJuergen Richter – Cognitec Systems GmbH
For legal reasons some pictures shown on the presentation at theEnd User meeting had to be removed or replaced for this version.
![Page 2: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/2.jpg)
2
Outline
• How does Cognitec's 3D Face Recognition work ?
• Resulting research tasks and achievementssince start of the EU 3DFace project
• Conclusions for development and user scenarios
![Page 3: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/3.jpg)
3
Face Recognition: Principal Approach
Raw Features
Input Data
Feature Vector
Data as provided by sensor
Features with reduced contingency
Features transformed to optimally distinguish individualities
Feature Extraction
Classification Transformation
![Page 4: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/4.jpg)
4
Characterization of Sensor Data
• Typically contains more than just the face• Varying orientation in space (pose)• Occasional data corruption
(gaps, outliers, artefacts like ripples)• Deformations of facial surface due to
glasses, hairdo and beard
![Page 5: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/5.jpg)
5
3D Face Recognition: Feature Extraction
•Localize facial part of 3D shape
•Align facial region to some defined orientation
•Preprocess shape to eliminate data flaws(Smoothing)
•Apply some feature extraction operator(Reduced dimension input for classification
![Page 6: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/6.jpg)
6
3D Face Recognition: Localization
• Very accurate 2D feature finders available
• Task of feature localization much more difficult for 3D shapes
=> Use Power of 2D feature finders for 3D localization, too!
![Page 7: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/7.jpg)
7
3D Face Recognition: Localization
Starting from 2D locations, find 3D locations bypixel – vertex correspondence
i = 1,2,...
k=1,2,...
![Page 8: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/8.jpg)
8
3D Face Recognition: Alignment
![Page 9: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/9.jpg)
9
3D Face Recognition: Shape Preprocessing
• Noise: deviations from true positions with some continuous random distribution
• Outliers: isolated points or groups of pointswith large deviation from true positions
• Gaps: locations where 3D data is missing at all• Occlusions: particular sort of gaps in regions not
available to sensor measurement
Eliminate data flaws related to:
One algorithm fits all:3D surface reconstruction by “Moving Least Squares”
![Page 10: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/10.jpg)
10
Least Squares Method (1)
Linear Regression:Find line linearly approximating a set of points=> minimize sum of squared distances
![Page 11: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/11.jpg)
11
Least Squares Method (2)
Disadvantage: Sensitive to outliers!
![Page 12: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/12.jpg)
12
3D Face Recognition: Moving Least Squares (MLS)
Local Least Squares Approximation
Probe points
MLS SurfaceTangential Plane
Local Neighbourhood
![Page 13: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/13.jpg)
13
3D Face Recognition: Processing Steps
Sensor Data MLS smoothedAligned
![Page 14: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/14.jpg)
14
3D Face Recognition: Major Research Tasks
• Improvement of 2D Feature Finders(face finder, eyes finder)
• Implementation and optimization of MLS algorithm
• Optimization of classification algorithm
“Improvement” and “Optimization” both in terms of biometric performance, robustness and computing speed!
![Page 15: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/15.jpg)
15
Selected Achievements: Face Finding
Face Finder Performance: % of faces not/incorrectly found
0.10.3Internal 30.11.7Weizmann
Strong variations in both pose and lighting:0.20.3Internal 20.050.2Internal 10.040.1Cferet-frontal
Mainly frontal images, low variation in lighting:T8 (2008)T6 (2006)Face Database
![Page 16: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/16.jpg)
16
Selected Achievements: Recognition Performance
![Page 17: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/17.jpg)
17
Selected Achievements: Computing Speed
0.40.9Most recent MLS versionFeb 2008
1.86.8MLS version in TDF ModulesDec 2007
n/a201st MLS version May 2006
4 Threads1 Thread
MLS Surface (200x200) Computation Timings (sec)on Xeon 3220 (QuadCore, 2.4 GHz)
![Page 18: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/18.jpg)
18
Avoidable Data Flaws
• Nonfrontal pose resulting in data gaps
• Occlusions caused by caps, dark glasses, strands of hair
• Data artefacts (ripples, gaps, outliers) caused by head motion during capture
![Page 19: Advances in Face Recognition...The face recognition company Advances in Face Recognition Research Presentation for the 2nd End User Group Meeting Juergen Richter – Cognitec Systems](https://reader034.vdocuments.us/reader034/viewer/2022042601/5f883686032a016b424b4a8f/html5/thumbnails/19.jpg)
19
Conclusions
• Keep exposure time low• Provide clear feedback on start and end of
exposure phase• Reduce processing times as far as possible
For user scenario: “Educate” cooperative user
For Sensor and Application Developers:
• Look straight into sensor camera• Keep neutral expression• For time of exposure, try to freeze head movements• Avoid occlusion of face (caps, dark sunglasses, hair)