An Automated Tool for Generating Monitoring Models for Complex Agrobiotechnical Systems

Abstract

This paper presents an implementation of an algorithm for the automatic structural-parametric synthesis of domain-specific models based on a modified genetic algorithm. The YOLO (You Only Look Once) architecture is used as the core framework, enabling high-speed image analysis on mobile devices. The proposed approach provides agricultural specialists with a tool for independently creating and deploying models and software solutions by uploading appropriate datasets to the AutoGenNet system for automatic neural network generation, without the involvement of machine learning experts. Verification and validation of the developed software were performed on a binary classification task using the example of vegetable crop condition assessment. The computational complexity of the proposed algorithm enables real-time evaluation of the state of a given class of objects.