Artificial Intelligence for hand gesture to speech conversion

This computer vision project entails building a web app that uses a computer&s;s web camera for capturing American Sign Language (ASL) gestures and converting them into audio and text simultaneously. This system is called HandTalk, which aims at solving the communication barrier by turning hand gestures into spoken words. In order to accomplish proper results in gesture recognition, the system implements two AI techniques, which are: Feedforward Neural Network (FNN) and Convolutional Neural Network (CNN). FNN, or Multi-Layer Perceptron (MLP), is particularly good at vision recognition and with sufficient training; it is able to notice incredibly accurate details of the subjects. Likewise, CNN is used to improve gesture recognition because it is very proficient in recognizing images. Based on the experiments conducted, the FNN classifier outperformed CNN with 98.23 percentage accuracy and 0.05 loss, while CNN&s;s figures were 96.6 percentage accuracy and 0.08 loss.