Gesture Recognition based on Deep Convolutional Neural Network

Abstract

A sign language is visually conveyed signal patterns to pass on meaning by concurrently coalescing hand shapes and direction and arms movement, body movement, facial expressions to articulate gracefully a narrator's feelings. Movement of the hand called hand gesture is the part of sign used by hearing-impaired inhabitants. It plays a key role for finger spelling in which signers spell out a word as a chain of hand shapes or hand flight equivalent to individual letters. Hand gesture recognition research classified as Glove based and vision based; Glove based system affects the natural signing ability. Vision based category is not using any devices or gloves, is the most natural way. Data acquisition can be done with camera. Conventional method of gesture recognition is time consuming so the usage of deep CNN is to prevent the segmentation or detection stage which favors faster real-time process.