Optimization of Gesture Recognition Algorithm-aid of Artificial Neural Netwok

Abstract

Objective:The Objective of our research is to establish an Artificial Neural Network (ANN) model for Braille coded gesture pattern recognition process and to optimize the weight of the ANN model using optimization algorithms. The ANN utilized to predict the outputs such as right gesture, left gesture, top and bottom gestures of the known input values. Methods: To optimize the weight of the ANN structure optimization techniques such as Genetic Algorithm (GA), Differential Evolution (DE), and Lion Algorithm (LA) are utilized to optimize the weights such as number of input neurons (α) and number of hidden neurons(β). Findings: The optimum results show that the attained empirical error values and predicted values are equal to zero in designed network. The Convergence graph demonstrates the fitness rate with minimum error 0.33 between different iterations for three optimization algorithms. The Lion algorithm results the optimal fitness value of Braille Coded Gesture ANN model. Applications: The lion algorithm can be utilized for gesture recognition in touch enabled devices in the area of human computer interaction.