Sign Language Recognition using Python and OpenCV

Sign Language Recognition using Python and OpenCV


Humans know each other by conveying their ideas, thoughts, and experiences to the people around them. There are numerous ways to achieve this and the best one among the rest is the gift of “Speech”. Through speech everyone can very convincingly transfer their thoughts and understand each other. It will be injustice if we ignore those who are deprived of this invaluable gift; the deaf and dumb people. The only means of communication available to the deaf and dumb people is the use of “Sign Language”. Using sign language, they are limited to their own world. This limitation prevents them from interacting with the outer world to share their feelings, creative ideas and Potentials. Very few people who are not themselves deaf and dumb ever learn to Sign language. These limitation increases the isolation of deaf and dumb people from the common society. Technology is one way to remove this hindrance and benefit these people.

The communication between a dumb and hearing person poses to be an important disadvantage compared to communication between blind and ancient visual people. This creates an extremely little house for them with communication being associate degree elementary aspect of human life. The blind people can speak freely by implies that of ancient language whereas the dumb have their own manual-visual language referred to as sign language. Sign language is also a non-verbal form of intercourse that's found among deaf communities at intervals the planet. The sign languages haven't got a typical origin and hence hard to interpret. A Dumb communication interpreter is also a tool that interprets the hand gestures to sensibility speech. A gesture in associate degree extremely language is also a certain movement of the hands with a particular kind created out of them. A gesture in a sign language is a particular movement of the hands with a specific shape made out of them. A sign language usually provides sign for whole words. It can also provide sign for letters to perform words that don’t have corresponding sign in that sign language. In this device Flex Sensor plays the major role, Flex sensors are sensors that change in resistance depending on the amount of bend on the sensor. This digital glove aims to lower this barrier in communication. It is electronic device that can translate Sign language into speech in order to make the communication take place between the mute communities with the general public possible. A hand gesture recognition system is also used to recognize real time gesture in unconstrained environments. The system consists of three modules: real time hand tracking, training gesture and gesture recognition using pseudo two-dimension hidden Markov models. In this they have used a Kalman filter and hand blobs analysis for hand tracking to obtain motion descriptors and hand region.

Sign language recognition (SLR) is an evolving research area in computer vision. The challenges in SLR are video trimming, sign extraction, sign video background modelling, sign feature representation and sign classification. All the problems are attempted in the past have met considerable amount of success and are instrumental in development of the state of the algorithms for SLR. Gesture recognition uses powerful imaging and artificial intelligence-based algorithms for classification. Current trends show an urge to bring gesture recognition into mobile environments. Sign language is visual mode of communication between two hearing impaired or hard hearing people. The communication foundations are based on finger shapes, hand shapes, hand movements in space with respect to body, hand orientations and facial expressions. The humans are trained exclusively to hand such huge amounts of information for years. For machine translation, the problem transforms into a 2D natural language processing problem. Many 1D/2D/3D models are proposed in literature with little success to bring the model close to real time implementation.

Sign language recognition (SLR) has transformed with technology upgradation from 1D, 2D to 3D models in the last 2 decades. In 1D, SLR is based on 1D signals acquired from a hand gloves and classified using signal processing methods. However, 1D methods are not up to the mark in terms of accuracy, and require hardware, complex circuitry, processor, etc. Also, the hand images are inherently 2 dimensional or 3 dimensional in nature. If we reduce it to one dimensional electrical signal, using some flex sensor, there will be a huge loss in data. With the development of modern methods like ML, DL, AI, ANN, CNN, etc. we can deal with higher dimensional image data directly through software like Python.

In this project the focus is to recognize the signs of American Sign language using the real time photos captured by the webcam.

The main advantage of CNN over other techniques like Random Forests, ML, ANN, etc. is that the CNN captures more number of features, and these are position invariant.



In these 2 figures, the locations of the stop sign are different. However, if we write a CNN to detect the presence of a stop sign, it will detect them. This may not be the case with other techniques.


Hence I and my team mate Shally Preethika Mani, have built a software, which can do this sign language classification.
Author: Sivasangaran V, Shally Preethika Mani
Project Guide: Dr B Vasuki


Click here for downloading this software.
Click here for viewing the user manual.

 

Comments