Contact less Palm print Authentication System

Project Title

Contact less Palm print Authentication System

Project Area of Specialization Artificial IntelligenceProject Summary

Biometrics features can be used for authentication purpose in computer-based security systems. The computers-based security systems are used at various places like commercial, civilian and government offices to store information and all processing. It is the primary thing to provide security to the information present on internet, for this purpose, the confidential authentication is required by replacing the username and password. The authentication systems on mobile devices do not have adequate security measures. The graphic patterns and passwords can be easily observed and stolen. Furthermore, the passwords are mostly short in length and therefore can be easily memorized. On the other hand, if people use lengthy and difficult passwords they can forget the passwords, and this can be harmful for the person and even for the mobile device. On contrary, the biometrics systems are quite reliable and have high security as they identify a user based on their physical characteristics. As the physical identities are barely lost or forgotten therefore the biometrics systems are considered better than the traditional authentication systems. Palm print images contain rich unique features for reliable human identification, which makes it a very competitive topic in biometric research.

In our project, we are using deep learning technique to train the model and get accurate and efficient results. For training the model we are using MobileNet architecture of convolutional neural network, which has given us the accuracy of 98.15%. MobileNet architecture is light weighted architecture and is thus reliable to deploy in mobile applications. It has 22 layers. We have trained our model by dividing the dataset into the ratio 90% and 10%, i.e. 90% training and 10% testing. We have used Python language for training and testing purposes and Java in development of android application. We have used Keras which is a neural network library with Tensorflow backend. We are targeting to develop an android application for contactless palm print recognition.

Project Objectives

The aim of this project is to implement, evaluate, and design deep learning techniques used for classification and recognition of user’s palms. We will try to provide and accurate, cheap, automatic and fast image processing-based solution for contactless palmprint authentication. Dataset is taken from a reliable internet source. Our project will be an android application. Convolution neural networks will be used for feature extraction from the images and for training the model. The keras pre-trained CNN model, the MobileNet architecture will be used in our project. The main objectives of project are:

• To extract features from palmprint images and training the model.
• To develop a user-friendly android application.
• To provide easier, effective, efficient, hygienic, low cost solution for palmprint authentication.
• To display the output as the label of the disease

The problem statement for this project is to develop a low-cost contactless palm print biometric system, as palm prints have rich unique features for reliable human identification which makes it competitive. The newest techniques of deep learning that we will be using are Convolution Neural Networks and Keras MobileNet architecture.

Project Implementation Method

we are using deep learning technique to train the model and get accurate and efficient results. For training the model we are using MobileNet architecture of convolutional neural network, which has given us the accuracy of 98.15%. MobileNet architecture is light weighted architecture and is thus reliable to deploy in mobile applications. It has 22 layers. We have trained our model by dividing the dataset into the ratio 90% and 10%, i.e. 90% training and 10% testing. We have used Python language for training and testing purposes and Java in development of android application. We have used Keras which is a neural network library with Tensorflow backend. Following tools are used:

• Miniconda 3.6.5
• Jupyter Notebook
• Tensorflow
• Keras
• Python & Java
• Convolution Neural Networks (CNN)
• MobileNet Architecture
• Model will be deployed into the android application.

Benefits of the Project

The computers-based security systems are used at various places like commercial, civilian and government offices to store information and all processing. It is the primary thing to provide security to the information present on internet, for this purpose, the confidential authentication is required by replacing the username and password. The authentication systems on mobile devices do not have adequate security measures. The graphic patterns and passwords can be easily observed and stolen. Furthermore, the passwords are mostly short in length and therefore can be easily memorized. On the other hand, if people use lengthy and difficult passwords they can forget the passwords, and this can be harmful for the person and even for the mobile device. On contrary, the biometrics systems are quite reliable and have high security as they identify a user based on their physical characteristics. As the physical identities are barely lost or forgotten therefore the biometrics systems are considered better than the traditional authentication systems. Palm print images contain rich unique features for reliable human identification, which makes it a very competitive topic in biometric research.

Technical Details of Final Deliverable

The biometric system should authenticate an administrator before giving them any type of access to the system.      

• The system after an assured number of unsuccessful authentication attempts should get locked.     
• The system must only allow the administrator to add, update or delete any type of data or any configuration parameters.
• The biometric authentication system should successfully recognize the users without any error.
• The system should ensure that the clients don’t get access to the administrator’s functions.
• The system should ensure that the access to the database is only available to the trusted people.
• The connection between the biometric recognition device, database and other components must be secured.
• The system must ensure that the user’s confidential data is properly protected.
• The system should be able to defend itself if an attack on the system or on the user information occurs.
• Any type of intrusion attack at least if not prevented must be detected by the system.
• The system should have intrusion detection and prevention software in it.
• When a virus attack or an unauthorized access attack is detected the system must as soon as possible inform the administrator about it.
• The system should never disclose any type of personal information of one user to any other users.
• The system should confirm that the image captured is of a living human being.
• The system must capture the images correctly and successfully. As the inputs are first and the most basic requirements for further procedures.
• At the first time, if due to user’s ignorance the image is not correctly captured the system should prompt a message for the user for recapturing the palm image.
• The system must extract the features from the captured images correctly for better authentication results.

Final Deliverable of the Project Software SystemType of Industry IT , Security Technologies Artificial Intelligence(AI)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	20000
GPU	Equipment	1	20000	20000