Development and Research of Hand Segmentation Algorithms on the Image Based on Convolutional Neural Networks

Abstract

In recent years, one of the most important scientific task in robotics is research aimed at developing methods that ensure a safe interaction between a robot and a human. An important aspect of solving this problem is the need to find a solution that allows obtaining maximum recognition accuracy. At the same time, the latest achievements in deep learning open up new possibilities in solving the task of object recognition. In this regard, the purpose of this work is to investigate the algorithms for recognizing human hands on the image, developed on the basis of deep convolutional neural networks.