Age and Gender Prediction

Project Title

Age and Gender Prediction

Project Area of Specialization Artificial IntelligenceProject Summary

Age and Gender are much important for the analysis of public behavior and for "shopping marts" to understand the tendency of people of different age and gender to purchase things. Once they analyze the behavior, they can change arrangements of product racks accordingly. In most of the "social media apps" and "web applications" children use the app providing fake age, using age prediction through image we can check whether the provided age is correct or not. In this project we will be providing a web interface where a user uploads a picture of face to check the age or gender and in result the user gets predicted age and gender. We propose the method of feature extraction from the extracted frame of a picture or video using techniques of principle component analysis (PCA) and then feed the extracted features to a layered convolutional neural network (CNN) to predict the age and gender of the given picture. In the project we will be using different datasets for training and testing like Adience-Face Benchmark face dataset,kaggle dataset and utkface dataset.

Project Objectives

1. Age prediction Using CNN that how old is the person

2. Gender Prediction Using CNN like (Male and Female)

To develop an web application that will allow end user to insert image and get estimated age and gender results of the inserted image.

Project Implementation Method

Gender Prediction: The gender is predicted using conv-2D layers based model. It contained 9 layers which includes pooling layers and output layers. The accuracy was about 95%. For age estimation we are using k-train model. The model will be able to predict estimated age for inserted picture. The interface of the project will be built in Flask. It will include an option to insert image, and options for age and gender prediction individually.

Age Prediction: 

We used resnet50 model to predict the age and trained it using Ktrain vision library. The model is using linear regression to get trained and in output it is providing the age of given picture. After that we made age groups like child, young, teenager, young adults, old adults and old etc. The final output of the project includes age, gender and age group of the given picture. For training of model the dataset from UTKFace was used. It has 23000 images thst include pictures of ages from 1 to 116. 10,000 images are Male and others are Female.

For the training and testing of the model GPU is used because of large dataset and layers of resnet50 model. The CPU takes very long time to train and test model on 23000 images.

Benefits of the Project

Gender detection is a very useful task for a wide range of applications. In the Spoken Language Systems, the Gender Detection module is one of the basic components of our audio segmentation system where it is used prior to speaker clustering, in order to avoid mixing speakers from different genders in the same cluster.

* Market basket analysis for different Age groups and Gender

* Public behavior analysis in public places.

* Check authentication of age in social media apps where age restrictions are applied.

Technical Details of Final Deliverable

The model of age and gender prediction is converted in to h5 file after training and testing. The web interface is developed in flask for third person usage. the users will be able to visit website and than will have option to insert image than after checking the requirements of image the output will bw shown. There is also an option for real time prediction of age and gender.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Security Core Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	66512
GeForce GTX 1660	Equipment	1	57812	57812
Cougar XTC500 Power Supply	Miscellaneous	1	8700	8700