Hand gesture recognition using a deep learning model

Abstract

Artificial neural networks called convolutional neural networks are modeled after visual cortex. CNN will be used to perform conversion of image to a value-matrix given to certain values than to others. As a output of this process, every neuron present in the layer is connected to a minimal portion of the layer that was available previously, without taking into account all of the neurons that are present in the fully connected network. To begin with, image samples collected so that gaussian blur and a threshold could be applied to them. The Image data generator function in Keras is used to perform image augmentation. Each image in the batch is subjected to a series of arbitrary translations, rotations, and other adjustments by ImageDataGenerator. In the proposed system, image features are extracted using the CNN algorithm.