MobileNetv2 Neural Network Model for Human Recognition and Identification in the Working Area of a Collaborative Robot

The article considers the software implementation of the MobileNetV2 neural network model for human recognition and identification in the working area of a collaborative robot. A mathematical description of the MobileNetV2 operation is presented, in particular its architecture and principles of operation, which allow to achieve high accuracy with reduced computing costs. The process of implementing the model in Python using the PyCharm environment is described, and a number of tests were conducted to evaluate its effectiveness in real-time conditions. The test results demonstrate the high accuracy and speed of the model, which confirms its suitability for use in collaborative robot systems that interact with people.