Human Face Localization in 3D For Humanoid Robot Vision using ML

Abstract

Robots, autonomous vehicles, security, and human-computer interaction are just a few of the many computer vision
applications that rely on accurate three-dimensional human position data. Improving the Arslan humanoid robot's
vision system so it can have a continual connection with humans in its field of view is the major emphasis of this
study. Using two identical USB cameras and two pan cameras calibrated with servo motors for precise angle
alignment, the system extracts depth information from photographs using human binocular vision principles. Taking
pictures, identifying faces, comparing camera counts, and doing 3D localization are all part of the process. The most
fundamental objectives are to maintain eye contact with people, to properly determine their exact positions, and to
predict their approximate locations. Recent advances in methodology have substantially improved localization
accuracy, leading to fewer errors and more reliability. The field of computer vision as a whole benefits from these
developments because they pave the way for more natural and quick interactions between humans and computers.