Evaluation of ensemble learning method for handwritten digits recognition using dual Leap Motions

Abstract

Recently, contactless input methods have been attracting attention. To meet such a demand, we focus on handwritten digits recognition. We install contactless hand tracking sensors on the lower and right sides of the subject's fingers and measure data from two directions for each subject’s handwritten digit. We analyze the three types of datasets composed of the data acquired by each sensor and the integrated data by using two types of machine learning models. Based on the results, we select combinations with high accuracy and construct an ensemble learning model. The classification accuracy achieves a maximum of 92.7%, applying the ensemble learning model with the integrated data.