<![CDATA[Child abduction is one of the most alarming social issues affecting communities worldwide. This study aims to develop an image-based child abduction detection and early warning system using a human object bounding box height ratio and distance comparison method. The proposed approach employs YOLOv8 object detection to identify adult and child human objects by analyzing and comparing the dimensions and spatial relationships of their bounding boxes. In this study, when two human objects are detected in close proximity and exhibit significantly different bounding box height ratios, the system classifies the situation as potentially suspicious, indicating a possible threat to a child playing in the monitored area. Consequently, the system generates an early warning notification and activates a buzzer alarm. Conversely, when one or more detected human objects are located at a relatively safe distance from each other, the system categorizes the condition as safe for the child. Experimental results demonstrate that the developed system is capable of providing early warnings when an adult approaches a child playing in a residential yard through both audible alarms and web-based notifications. Furthermore, parents or guardians can access information detected by the system, including anomaly records in the form of captured images and detailed information regarding the suspected perpetrator and the potential victim.]]>
Copyrights © 2026
