Dynamic Hand Gesture Pattern Recognition Using Probabilistic Neural Network

Flex Sensor and Gyroscope based hand gloves are being used widely to detect gesture-based sign language patterns to provide aid for speech-impaired people. However, detecting dynamic gestures is not an easy task due to the variability and redundant implementation precautions. This article approaches a dynamic hand gesture recognition system by using Probabilistic Neural Network (PNN) deliberately focused on practical usage of dynamic gestures on day-to-day life. In this experiment, 10 gesture patterns have generated by using a data glove. For each gesture, 360 training input vectors are generated to train the PNN model, and the output has provided via a speaker.