Аналіз підходів до програмної реалізації систем аналізу міських сцен з використанням 3w-класифікації ризику

Abstract

This paper analyzes approaches to the software implementation of urban scene analysis systems using 3W risk classification. The limitations of object-oriented solutions for risk assessment in complex scenes are demonstrated, and the feasibility of semantic-oriented and prototype-oriented approaches is substantiated. The possibility of applying three-way classification to handle uncertain situations and improve the interpretability of results is considered. Based on the conducted analysis, a combined approach is proposed: object-oriented recognition at the lower level for speed, and prototype-oriented risk assessment at the higher level, supported by three-way decisions for result interpretability.