Sign inSign up
Challenges in Learning the Appearance of Faces for Automated Image Analysis - Part 2 thumbnail
slide-image
Pause
Mute
Subtitles not available
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Challenges in Learning the Appearance of Faces for Automated Image Analysis - Part 2

Published on Feb 25, 20073625 Views

Thomas Vetter

The variability of images of the human face challenges research in machine vision since its beginning. Sources of variability not only include individual appearance but also cover external parameters

Workshop on Pattern Recognition and ML 2004 - Grenoble

Chapter list

Learning the appearance of faces for face recognition applications Pose and Illumination Invariant Face Recognition00:22
The Problem01:14
Pose and Illumination Variation01:46
Face Identification by Image Comparison03:20
3D Shape + Illumination Inversion03:53
Analysis by Synthesis04:46
Image Synthesis05:37
Learning Image Models from Examples06:00
Approach: Example based modeling of faces06:45
Representing Shape - 2D or 3D ?07:16
Outline08:09
3D Morphable Model Construction09:17
Cylindrical Coordinates09:42
Morphing 3D Faces10:48
Shape and Texture Vectors12:28
Shape and Texture Vectors12:57
Vector space of 3D faces.13:20
Active Appearance Model13:57
Dynamic Link Architecture14:36
Continuous Modeling in Face Space15:34
Image Rendering16:37
3D Morphable Model = ?17:26
Image Formation Process18:43
Vertex Projection22:07
Triangle List22:52
Surface Rendering23:02
Shape representation of 3DMM and AAM23:35
Triangle List24:31
Face Identification by recovering shape, albedo, lighting and pose from a single photographs24:32
Identification from Model Coefficients24:52
Fitting a Morphable 3D-Face-Model25:46
Energy Function26:42
Energy Function27:56
Error Function – MAP Estimate28:57
Derivatives30:38
3DMM-SNO – Fitting results30:48
Correct Identification “1 out of 68” (%)32:22
3DMM-SNO33:45
Different Fitting Algorithms34:30
Different Fitting Algorithms35:03
AAM Fitting35:47
AAM Fitting36:01
AAM Fitting a.k.a. IDD36:48
AAM - Identification37:34
Inverse Compositional Image Alignment39:04
ICIA – Example40:13
ICIA – Example40:39
ICIA applied to the 3DMM42:25
ICIA applied to the 3DMM42:29
3DMM-ICIA – Fitting Results42:48
Image Preprocessing for FRVT 200243:34
Image Preprocessing for FRVT 200244:37
Image Preprocessing for FRVT 200245:13
References46:20
Question one47:05
Question two49:36
Big Question52:18