Abstract:
Problem: Children may interact with inappropriate or harmful objects without having parents’
awareness. Also, current education applications have a lack of real time filtering system of
object detection with age appropriateness. This project addresses an AR study game that
identifies real time objects and filters that identify objects based on age properness.
Methodology: Here Project use YOLOv8 for real-time object detection and filtering model
and the object filtering model is learned with a supervised learning technique using the detected
age group. The AR system displays educational content that is engaging, and questions. Added
AR feedback which enables children to collect information on the perceived objects by actively
answering questions.
Initial Results: The first results prove that the YOLOv8 model accurately detects objects. The
CNN model provides 63.32% accuracy within a range of ±5 years of age. Therefore, age-based
content filtering is possible. The integrated system effectively provides the interactive
educational prompts and feedback with well-designed UI activities, yet they require
improvements in the network response time.
Subject Descriptors:
• Computing methodologies -> Machine learning -> computer vision -> real time object
detection
• Human computer interaction -> augmented reality
• Computing methodologies -> Machine learning -> Data mining -> objects filtering
algorithm
