With the rapid development of AI technology, Face recognition system, and Speech recognition system, as new systems combining computer image processing technology and biometric technology, have attracted people's attention and made people more and more interested in it.
This project is dedicated to a face & speech recognition system that can be used for airport security. A passenger only need to look at the camera and say his/her own name to microphone, then the access gate is open and boarding record is saved.
The face & speech recognition system can reduce the verification time and improve the accuracy of comparison. It have outstanding contribution in protecting security and tracking fugitives.
ml5.js aims to make machine learning approachable for a broad audience of artists, creative coders, and students. The library provides access to machine learning algorithms and models in the browser, building on top of TensorFlow.js with no other external dependencies.
A class of efficient models called MobileNet for mobile and embedded vision applications. MobileNet is based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks. The effectiveness of MobileNet across a wide range of applications is high and its use cases include object detection, finegrain classification, face attributes and large scale geo-localization.
Photos collected should be taken at different times, with different lighting, different expressions (open eyes and closed eyes, laugh or not), and different face details (with or without glasses).
The more photos collected, the higher face recognition accuracy of the system.
Training AI is essential.
Machine learning has an ability to identify patterns. The code below involves image classification (video) technology.
Firstly, collect voiceprint information and establish biometric library: voiceprint database.
The voiceprint collection system is mainly composed of two parts, one is the hardware-based voiceprint collector, and the other is the software-based voiceprint collection management system. The voiceprint collector is mainly responsible for the collection of high-fidelity sound to ensure that the collected sound files It can also be used in the process of subsequent iterations of the algorithm and hardware.
Also, training AI is essentail.
Based on the boimetic library, the code below provides the method to implement speech recognition.
Nowadays, face recognition is nothing new now. Face payment and face-access control have gradually entered people's lives. Siri, Google Assistant and other speech recognition systems also provide convenience for people's daily lives.
However, how to properly use AI technology and how to protect people's privacy is worth pondering.
