ICTACT Journal on Soft Computing

Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Deep Machine Learning for Age and Gender Prediction


Affiliations
1 Department of Electrical and Information Engineering, University of Nairobi, Kenya
     

   Subscribe/Renew Journal


This paper tends to show that by learning feature representations through the employment of convolutional neural networks (CNN), a major increase in performance is obtained on age and gender prediction tasks. An image classifier is built in Matlab. Thousands of facial images are obtained and used to train a convolutional neural network. In this case of deep learning, the CNN essentially constructs abstract features from training image data, which would otherwise have to be handcraft in traditional machine learning model. Feeding in an image in an input, each layer it will perform a series of operations on that data until it outputs a label and classification percentage. Each layer has a different set of abstractions; in the first layers, the network basically teach itself edge detection, then shape detection in the middle layers. They get increasingly more abstract up until the end. The last few layers are the highest-level detectors for the whole object. A lot of computing power and time is spent to train the deep network. The trained network is then used to do predictions of age and gender and can, later after this paper, be integrated with webcam, at home or office to get statistical summary of all guests’ age and gender.

Keywords

Age, Gender, Prediction, Convolutional Neural Networks, Deep Machine Learning.
Subscription Login to verify subscription
User
Notifications
Font Size

  • N.S. Chauhan, “Predict Age and Gender using Convolutional Neural Network and openCV”, Available at: https://towardsdatascience.com/predict-age-and-gender-using-convolutional-neural-network-and-opencv-fd90390e3ce6.

  • Historian, “Marvin Minsky”, Available at: https://www.i-programmer.info/history/people/368-marvin-minsky.html.

  • Dong-Phuong Nguyen, Rudolf Berend Trieschnigg, A. Seza Dogruoz, Rilana Gravel, Mariet Theune, Theo Meder and Franciska M.G. De Jong, “Why Gender and Age Prediction from Tweets is Hard: Lessons from a Crowdsourcing Experiment”, Proceedings of 25th International Conference on Computational Linguistics, pp. 1950-1961, 2014.

  • Eran Eidinger, Roee Enbar and Tal Hassner, “Age and Gender Estimation of Unfiltered Faces”, IEEE Transactions on Information Forensics and Security, Vol. 9, No. 12, pp. 2170-2179, 2014.

  • G. Levi and T. Hassner, “Age and Gender Classification using Convolutional Neural Networks”, Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition Workshops, pp. 1-6, 2015.

  • A.C. Gallagher and T. Chen, “Clothing Cosegmentation for Recognizing People”, Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition, pp. 23-28, 2015.

  • Sebastian Lapuschkin, Alexander Binder, Klaus-Robert Muller and Wojciech Samek, “Understanding and Comparing Deep Neural Networks for Age and Gender Classification”, Proceedings of IEEE International Workshop on Analysis and Modeling of Faces and Gestures, pp. 1-8, 2017.

  • S. Escalera, J. Fabian, P. Pardo, X. Baro, J. Gonzalez and H.J. Escalante, “ChaLearn Looking at People 2015: Apparent Age and Cultural Event Recognition Datasets and Results”, Proceedings of IEEE International Conference on Computer Vision Workshop, pp. 7-13, 2015.

  • Rasmus Rothe, Radu Timofte and Luc Van Gool, “IMDB-WIKI-500k+ Face Images with Age and Gender Labels”, Available at: https://data.vision.ee.ethz.ch/cvl/rrothe/imdb-wiki/.

  • P. Skalski, “Gentle Dive into Math behind Convolutional Neural Networks, Mysteries of Neural Networks Part V”, Available at: https://towardsdatascience.com/gentle-dive-into-math-behind-convolutional-neural-networks-79a07dd44cf9.

  • R. Rothe, R. Timofte and L.V. Gool, “Deep Expectation of Real and Apparent Age from a Single Image without Facial Landmarks”, International Journal of Computer Vision, Vol. 126, No. 2-4, pp. 144-157, 2016.

  • R. Rothe, R. Timofte and L.V. Gool, “DEX: Deep Expectation of Apparent Age from a Single Image”, Proceedings of IEEE International Conference on Computer Vision Workshops, pp. 121-127, 2015.

  • E. Agustsson, R. Timofte, S. Escalera, X. Baro, I. Guyon and R. Rothe, “Apparent and Real Age Estimation in Still Images with Deep Residual Regressors on APPA-REAL Database”, Proceedings of IEEE International Conference on Automatic Face and Gesture Recognition, pp. 30-36, 2017.

  • Y. Uchida, “APPA Real Dataset”, Available at: https://github.com/yu4u/age-gender-estimation/tree/master/appa-real.

  • K. Ravishankar, “Age and Gender Classification using MobileNets”, Available at: https://medium.com/ymedialabs-innovation/estimating-age-and-gender-with-mobilenets-13eaee1e819c.

  • Precision and Recall, Available at: https://en.wikipedia.org/wiki/Precision_and_recall.

  • C.M. Bishop, “Neural Networks for Pattern Recognition”, Oxford University Press, 2005.


Abstract Views: 894

PDF Views: 0




  • Deep Machine Learning for Age and Gender Prediction

Abstract Views: 894  |  PDF Views: 0

Authors

George Wanga
Department of Electrical and Information Engineering, University of Nairobi, Kenya
Segera Rene Davies
Department of Electrical and Information Engineering, University of Nairobi, Kenya

Abstract


This paper tends to show that by learning feature representations through the employment of convolutional neural networks (CNN), a major increase in performance is obtained on age and gender prediction tasks. An image classifier is built in Matlab. Thousands of facial images are obtained and used to train a convolutional neural network. In this case of deep learning, the CNN essentially constructs abstract features from training image data, which would otherwise have to be handcraft in traditional machine learning model. Feeding in an image in an input, each layer it will perform a series of operations on that data until it outputs a label and classification percentage. Each layer has a different set of abstractions; in the first layers, the network basically teach itself edge detection, then shape detection in the middle layers. They get increasingly more abstract up until the end. The last few layers are the highest-level detectors for the whole object. A lot of computing power and time is spent to train the deep network. The trained network is then used to do predictions of age and gender and can, later after this paper, be integrated with webcam, at home or office to get statistical summary of all guests’ age and gender.

Keywords


Age, Gender, Prediction, Convolutional Neural Networks, Deep Machine Learning.

References