Docker

The paper presents the development of a cyber-physical system based on the HomeAssistant platform for efficient automation and control of Internet of Things (IoT) devices. The architectural features, technical implementation and prospects for the development of the system are considered, with an emphasis on the ability to integrate a wide range of sensors and IoT devices into a single network to create adaptive and intelligent solutions.

As a result of the analytical review, it was established that the family of Yolo models is a promising area of search and recognition of objects. However, existing implementations do not support the ability to run the model on the iOS platform. To achieve these goals, a comprehensive scalable conversion system has been developed to improve the recognition accuracy of arbitrary models based on the Docker system. The method of improvement is to add a layer with the Mish activation function to the original model. The method of conversion is to quickly convert any Yolo model to CoreML format.