Комп’ютерний зір для контролю безпеки ігрового середовища дитини

Abstract

This research addresses the challenge of ensuring child safety in domestic environments using image analysis and computer vision. The paper proposes an intelligent system that analyzes static images of a child’s play area to detect both the child and potentially dangerous nearby objects. The approach is based on convolutional neural networks and the YOLOv10 object detection architecture. Special attention is given to spatial proximity analysis and automated risk estimation by calculating the Euclidean distance between detected objects to assess potential threats. Experimental results demonstrate stable detection accuracy under typical indoor conditions. The proposed solution is intended to support parents by providing timely alerts and reducing everyday domestic risks.