Age recoginition via fastAAMs
A very simple approach!
The code is in the
It includes two steps:
train AAM Model
% featureExtraction.m % the training data is in `morph_small/trainset/` flag_train = 1; flag_precalibration == 1
flag_train = 1;, then
AAM.matfile (AAM model) will be generated in the folder
morph_small/trainset/. After that, set
flag_train = 0;.
flag_precalibration == 1;, then the calibration parameters would be computed and saved in
meantrans.mat. After that, set
flag_precalibration = 0;.
extract the training images using the AAM model
The training/test images are in the folder
morph_small/Images_ori/ (total 2500 images).
dataset the morph dataset can be download from this website.
- the piecewirse affine warpped images are generated in the folder
- the images with extracted features are generated in the folder
- and the extracted batch features in the
.matform are saved in the folder
collect all extracted features in a single
If the images dataset is very large, it may take a long time to train.
So you can partition the datasets into multiple parts, and train each part separately. Just remember rename the destination folder, like
Then run the
This script will collect all features from batch features into a single
append_features.m, the features are generated in
main.m train and test a classification model using
is_partition_dataset == 1 when first runs to generate
is_svmtrain == 1 and set
svm_type= 0 or 1 or 2 to train a svr model.
is_svmtest == 1 to test the model.
test_images.m script predicts the age from images.
REPORT is given: age_estimation_report.pdf
 Xin Geng, Zhi-Hua Zhou, and Kate Smith-Miles. Automatic age estimation based on facial aging patterns. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 29(12):2234–2240,2007.
 Georgios Tzimiropoulos and Maja Pantic. Optimization problems for fast aam fitting in-thewild. In Computer Vision (ICCV), 2013 IEEE International Conference on, pages 593–600.IEEE, 2013. http://www.mathworks.com/matlabcentral/fileexchange/44651-active-appearance-models--aams-
 Khoa Luu, Karl Ricanek, Tien D Bui, and Ching Y Suen. Age estimation using active appearance models and support vector machine regression. In Biometrics: Theory, Applications, and Systems, 2009. BTAS’09. IEEE 3rd International Conference on, pages 1–5.IEEE, 2009.
 Ricanek, Karl, and Tamirat Tesafaye. "Morph: A longitudinal image database of normal adult age-progression." Automatic Face and Gesture Recognition, 2006. FGR 2006. 7th International Conference on. IEEE, 2006.