Sign Language Learning System with Image Sampling and Convolutional Neural Network
Yangho Ji, Sunmok Kim, Ki-Baek Lee · 2017
This paper proposes a novel sign language learning system based on 2D image sampling and concatenating to solve the problems of conventional sign recognition. The system constructs the training data by sampling and concatenating from a sign language demonstration video at a certain sampling rate. The learning process is implemented with a well-known network, convolutional neural network. 6 sign language actions are learned in 20 situations and then the system is tested in an uninformed situation. The results show that the system is accurate and robust even if only 2D images obtained with low-cost cameras are used.