AUGMENTED REALITY-BASED ALPHABET LEARNING WITH REAL-TIME OBJECT DETECTION

Abstract

In the current changing digital learning environment, conventional approaches to early childhood education lacks the ability to maintain young learner's interest. Currently, the resources like flashcards and books do not link abstract ideas, like alphabet letters, to a child's daily surroundings. The Augmented Reality Alphabet Learning tackles this issue by providing an engaging, interactive environment where kids can discover and learn letters through the recognition of actual objects. This solution combines AR and deep learning to enhance education, making it more engaging and intuitive particularly in settings with restricted internet connectivity

The Research detects physical items via a DenseNet121-based object detection model and presents matching 3D models with audio narration through AR. Technologies such as TensorFlow Lite, ARCore, React Native, and Blender 3D are utilized for smooth object identification and content presentation. Important aspects comprise immediate object identification, three-dimensional animation, speech synthesis narration, assessments, progress monitoring, and controls for parents.

To verify the system's effectiveness, comprehensive testing was conducted that encompassed unit, UI, scalability, error-management, and user acceptance testing. Findings indicated seamless object detection, rapid AR rendering, and precise progress reporting on various Android devices. The app exhibited consistent responsiveness during prolonged usage and received favorable ratings for its user-friendliness, particularly from kids and non-technical individuals. Database testing verified correct monitoring of quiz outcomes and performance metrics.

Even with its impressive performance, there are also some limitations to the AR Alphabet Learning Research. Object detection can be impacted by poor lighting, complex backgrounds, and object orientation, sometimes leading to inaccurate results. Device performance differences affect AR quality, and current support is limited to English and a small set of recognizable objects. In future, we aim to includes expanding language support, improving detection accuracy with advanced algorithms and increasing training data set, enhancing AR with gamified features, and introducing cloud-based tracking for better progress monitoring.