Signify - A Sign Language Detection

The Idea Of “Signify” A Sign Language Detection Using Image Processing. In this project, we present a technique to detect American Sign Language (ASL) from images that performs in real time.

Objective

• The Objective Of This Project Is To Identify The Symbolic Expression Through Images So That The Communication Gap Between A Normal And Hearing Impaired Person Can Be Easily Bridged.
• To Develop An Automatic Sign Language Detection System With The Help Of Image Processing And Computer Vision Techniques.
• Communication Is Always Having A Great Impact In Every Domain And How It Is Considered The Meaning Of The Thoughts And Expressions That Attract The Researchers To Bridge This Gap For Every Living Being.
• To Provide A Real Time User Interface So That Signers Can Easily And Quickly Communicate With Non-Signers.
• To Efficiently And Accurately Recognize Signed Words, From ASL, Using A Minimal Number Of Training

Device

Webcam is used for the object detection.

Project Features

• Detecting Sign Language From Human Pose Estimation.
• Labeling New Images.
• Train Images For Sign Language.
• Real Time Sign Language Recognition Using Image Processing.
• Hand Gesture Recognition For Sign Language.
• Finger Detection For Sign Language Recognition.
• OpenCV For Faster Image Processing.
• Use of TensorFlow give flexibility and control with feature.

How It Works

First of all, we start with collecting the images for deep learning using our webcam and OpenCV. We take a handsome amount of pictures of each class of signs. Then we label the images for sign language detection using LabelImg. To do that, we crop out the irrelevant region to remove the noise, mark only the portion of the image containing the sign relevant to us, and then label the image. This will produce an label map file containing the annotations.We then create the record file to fix how many signs we’re going to train our system with as well as their order. It creates record files for both training and testing.After training the system with the datasets, it is then time to test it.

Tutorial : Youtube Link

Screenshorts

Call Sign

Nice Sign

Start Sign

Stop Sign

Victory Sign

Future Work

Future work will be extended for further improvement in recognition accuracy and also for movement detection of hand for word recognition. We will try to detect alphabets , numbers also.