Face Recognition on the MORPH-II Database Face Recognition on the MORPH-II Database Morgan Ferguson University of North Carolina at Wilmington July 25, 2017 1 / 26
Face Recognition on the MORPH-II Database Overview Introduction/ Background Morph-II Database Face Recognition Our Work with Face Recognition Goals Methods Results Analysis Conclusion Future Research References 2 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Morph-II Database Morph-II Database ◮ Contains 55,134 mugshots of 13,617 individuals ◮ Collected over 5 years ◮ Ages range from 16 to 77 years ◮ Average of around 4 pictures per individual ◮ Provides race, gender, date of birth, date of arrest, age, age difference since last picture, subject identifier, and picture number for each picture in the database 3 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Morph-II Database Challenges and Improvements ◮ Original data yields poor accuracy rates ◮ Images have different sizes and faces are in different locations ◮ Pre-processed images! ◮ Standardized data Original Modified 4 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Face Recognition What is face recognition? ◮ Process of identifying a new face as a known individual or unknown individual ◮ Face recognition works by training some classifier on a set of images (training or gallery) and then matching new image (test or validation) Figure: http://www.nec.com/en/global/solutions/safety/face_recognition/index.html 5 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Face Recognition Eigenfaces ◮ Face images are projected onto a feature space (”face space”) ◮ Face space is defined by ”eigenfaces”, or the eigenvectors of the set of faces ◮ Ability to learn to recognize new faces in unsupervised manner Figure: M. Turk, A. Pentland, ”Eigenfaces for Recognition”, J. Cognitive Neuroscience, vol. 3, no. 1, 1991. 6 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Face Recognition Fisherfaces ◮ Linear Discriminant Analysis (LDA) ◮ Maximizes distance between classes ◮ Supervised technique Figure: http://www.scholarpedia.org/article/Fisherfaces 7 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Face Recognition Local Binary Patterns (LBP) Review ◮ Different approach to obtaining vectors from an image ◮ Labels pixels of an image by analyzing the neighbors of each pixel and considers the result as a binary number ◮ LBP’s have parameters such as block size and radius Figure: http://www.scholarpedia.org/article/Local_Binary_Patterns 8 / 26
Face Recognition on the MORPH-II Database Introduction/ Background Face Recognition Classification and Distance Metrics Support Vector Machine (SVM) ◮ Radial basis Nearest Neighbor approach ◮ Euclidean distance ◮ Cosine distance ◮ Cityblock distance Figure: http://www.researchgate.net/figure/5423571_fig6_ Figure-13-313-Illustration-of-distance-measures ◮ Bray-Curtis distance ◮ Canberra distance ◮ Mahalanobis Cosine distance 9 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Goals Goals Reproduce Results ◮ Compare benchmark results with previous research New Trials ◮ Test out new subsets, distance metrics, feature vectors, etc. Optimize ◮ Try out different methods to improve on initial results Analyze ◮ Start creating tables and graphs to compare results ◮ Begin to make conclusions 10 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Methods Experimental Setup 1. Subset Morph-II 2. Begin with input data (pre-processed images from subsets or LBP feature vectors) 3. Break into training and testing data 4. Perform PCA (Eigenfaces) or LDA (Fisherfaces) 5. Use SVM or Nearest Neighbor 6. Classify test image as a subject from training images 11 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Methods Subset Scheme 1 ◮ Take subjects with more than 10 images each ◮ Randomly select 10 images for each person ◮ Match the number of males with the number of females (83 subjects each) ◮ Randomly select 1 image for each subject as testing image and designate other 9 for training ◮ 166 subjects and 1660 total images 12 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Methods Subset Scheme 2 ◮ Take subjects with more than 10 images each ◮ Randomly select 10 images for each person ◮ Randomly select 5 images for each subject as testing image and designate other 5 for training ◮ 544 subjects and 5440 total images 13 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results Initial Subset 1 Results Table: Accuracy Rates and Run Times for Face Recognition Algorithms (before histogram equalization) SVM-R Euclidean CityBlock Cosine Eigenfaces Accuracy 74.1% 50.6% 66.9% 59.6% (PCA) Run Time 177.3 5.2 3.9 8.6 (sec) Fisherfaces Accuracy 86.14% 87.3% 83.1% 95.8% (LDA) Run Time 165.2 7.7 6.1 13.3 (sec) 14 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results More Subset 1 Results Table: Accuracy Rates and Run Times for Face Recognition Algorithms (after histogram equalization) SVM-R Euclidean CityBlock Cosine Eigenfaces Accuracy 88.6% 69.9% 78.9% 71.1% (PCA) Run Time 211.3 7.4 3.9 10.0 (sec) Fisherfaces Accuracy 89.2% 92.8% 89.2% 95.8% (LDA) Run Time 179.4 10.8 7.3 12.8 (sec) 15 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results LBP Results (PCA) 16 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results LBP Results (LDA) 17 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results Overall Subset 1 Results Table: Face Recognition Accuracy Rates on MorphII Subset 1: Train on 9, Test on 1 SVM-R (%) Euclidean (%) CityBlock (%) Cosine (%) Eigenfaces 88.6 69.9 78.9 71.1 (PCA) Fisherfaces 89.2 92.8 89.2 95.8 (LDA) 87.3 71.7 69.3 76.5 LBP + PCA (s=10, r=1) (s=10, r=1) (s=12, r=1) (s=10, r=1) 88.6 84.9 81.3 89.2 LBP + LDA (s=10, r=2) (s=14, r=2) (s=14, r=2) (s=14, r=3) 18 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results Back to the Data Truth Prediction 19 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results Subset 2 Results Table: Accuracy Rates and Run Times for Face Recognition Algorithms using 5 training images, 5 testing Euclidean CityBlock Cosine BrayCurtis Canberra Eigenfaces Accuracy (%) 54.0 63.1 56.0 69.7 66.0 (PCA) Run Time 139.2 78.8 268.5 121.6 223.9 (sec) Fisherfaces Accuracy (%) 62.9 55.9 78.2 71.7 51.9 (LDA) Run Time 163.7 115.3 281.0 146.1 255.8 (sec) 20 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Results Subset 1 vs. Subset 2...What Happened? SVM-R Euclidean CityBlock Cosine BrayCurtis Canberra Eigenfaces Accuracy (%) 88.6 69.9 78.9 71.1 79.5 79.5 (PCA) Run Time 211.3 7.4 3.9 10.0 6.8 9.6 (sec) Fisherfaces Accuracy (%) 89.2 92.8 89.2 95.8 94.8 83.1 (LDA) Run Time 179.4 10.8 7.3 12.8 8.5 13.3 (sec) Euclidean CityBlock Cosine BrayCurtis Canberra Eigenfaces Accuracy (%) 54.0 63.1 56.0 69.7 66.0 (PCA) Run Time 139.2 78.8 268.5 121.6 223.9 (sec) Fisherfaces Accuracy (%) 62.9 55.9 78.2 71.7 51.9 (LDA) Run Time 163.7 115.3 281.0 146.1 255.8 (sec) 21 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Analysis Size Analysis 22 / 26
Face Recognition on the MORPH-II Database Our Work with Face Recognition Analysis Gender Analysis 23 / 26
Face Recognition on the MORPH-II Database Conclusion Future Research Future Research ◮ How can we optimize the face recognition algorithm further? ◮ How does race affect face recognition? ◮ How much would gender and race classification as a first step increase accuracy rates and reduce run time? 24 / 26
Face Recognition on the MORPH-II Database Conclusion References References G. Guo, G. Mu, K. Ricanek (2010) Cross-age face recognition on a very large database: the performance versus age intervals and improvement using soft biometric traits 20th International Conference on Pattern Recognition 2010 , 3392-3395. DK Hayati PHM Yassin, S.Hoque and F. Deravi (2013) Age sensitivity of face recognition algorithms Proc of the Fourth International Conference on Emerging Security Technologies (EST) , 2013 M. Turk, A. Pentland (1991) ”Eigenfaces for Recognition” J. Cognitive Neuroscience , vol. 3, no. 1, 1991. K. Ricanek Jr. and T. Tesafaye (2006) ”MORPH: A Longitudinal Database of Normal Age-Adult Progression” IEEE 7th International Conference on Automatic Face and Gesture Recognition (FGR06) , Southampton, UK, Apr. 2006, pp. 341345. 25 / 26
Face Recognition on the MORPH-II Database Conclusion References The End 26 / 26
Recommend
More recommend
