Deep Learning Algorithms for Secure Robot Face Recognition in Cloud Environments

摘要

To improve human-machine interaction in real-time robotic applications, especially in face recognition task, speed is limited because of robot on board limitations. The speed can be improved by incorporating the cloud technology, it supports flexibility in throughput, less cost, usability enhancement and fewer training instruction, high speed, but private cloud has security issues. So in this paper an attempt is made to improve speed and accuracy of recognition in encrypted domain without loosing privacy of robot. An overview of security algorithms like Cryptography based and image processing based schemes are presented. Some experiments are conducted for robot face recognition through various deep learning algorithms after encrypting the images (ORL database) and experimental results shows improvement in speed and accuracy of recognition.