[4Rin1] Interactive 2

[4Rin1-51] A Keyless System Combining Deep Learning-based Multi-Person Matching and Smart Lock Technology

〇Ryosuke Tanno¹, Shuhei Asano¹, Satoki Shibata¹, Masaya Okamoto¹, Yuri Ohata², Haruto Sakamoto², Masayuki Asanuma³, Koichiro Kawamura³ (1.NTT Communications Corporation, 2.NTT Worldwide Telecommunications, 3.KOZO KEIKAKU ENGINEERING Inc.)

Keywords:Person Re-Identification, Face Recognition, Body Recognition, Pose Estimation, Smart Lock

In recent years, demand for the use of data centers has increased due to the development of cloud technologies, and the data center market has been expanding year by year. In data centers where high security is required, key management is performed for each server rack, and in large-scale data centers, as many as 20,000 keys are lent annually. It is estimated that it takes more than 5000 hours a year to respond to the loaning of keys, and there are problems such as increased waiting time for data center reception and increased operation. Therefore, in this study, we propose a keyless data center that combines deep learning-based authentication technology using human skeleton information and electronic lock technology "RemoteLock" provided by the Structural Planning Research Institute, in addition to commonly used human verification technology such as face authentication and whole-body authentication.