Deepface

Facebook's Face Recognition system

1. DeepFace:
Closing the gap to human level
performance in face verification
Written by:
YanivTaigman, Ming Yang , Marc’ Aurelio, Loir Wolf
Presenter:
Salu Khadka
Trinity International College
[email protected]
Mark Zuckerberg

2. About the paper……
•YanivTaigman, Ming Yang, Marc'AurelioRanzato, LiorWolf
•(June 24, 2014) .
•Conference on Computer Vision and Pattern Recognition
(CVPR)
•To reduce the error of the current state of the other
systems by more than 27%, closely approaching human-
level performance.

3. FACE RECOGNITION SYSTEM
•Era of Big Data
•A computer application for automatically
identifying or verifying person from a digital
image or a video source.
•Highly applicable in security and access
control, biometrics , entertainment etc.

4. PROBLEM IN FRS
A-PIE
EMOTION
ILLUMINATION POSE
AGING
PIE

5. DEEPFACE
DETECT ALIGN REPRESENT CLASSIFY
•Appearance-based holistic approaches
•Deep learning
•Accuracy of 97.35%
PIPELINE OF RECOGNITION SYSTEM

6. FACE ALIGNMENT
2D representation
6 facial point
Translate
Scale
Rotate
Out of plane pictures
A-PIE
Fig: Generic 3D model

10. REPRESENTATION
Convolutional neural network
•feed-forward artificial neural network
•minimal amount of pre processing.
•easier to train.

11. Extracts low level features Lessens the parameter

12. FEEDBACK NETWORK

13. CONCLUSION
•Made proper use of deep learning.
•Reduced the error rate of other system.
•Contributes in visual data .