Improved IoT-Based Child Detection System using Enhanced Behavioural Analysis and Noise-Filtered KNN

In recent years, many parents face increasing challenges in continuously monitoring their children's activities. Although the Internet of Things (IoT) supports early detection and monitoring, its reliability is often limited due to the shortcomings of existing methods, such as K-Nearest Neighbour (KNN). To solve this, the proposed method uses Edited Nearest Neighbours (ENN) to remove duplicate and irrelevant data, optimizing preprocessing and improving accuracy. Additionally, Improved Behavioral Analysis using Deep Learning (IBADL) examines both positive and negative behaviours, providing a comprehensive overview for improved monitoring. This approach focuses on early prevention to reduce children's exposure to harmful elements and monitors both indoor and outdoor activities. The video input detects nearby objects, recognizes children, and monitors the surroundings, and measures activity performance using tested training data. This system enhances reliability, ensures scalability, and reduces time complexity while maintaining an accuracy of 91%. The combined techniques help parents gain valuable insights and enhance children's safety.