Background and Motivation

Gap 1: Cantonese-Based Learning

Most market tools prioritize Mandarin or English. This project emphasizes locally used Hong Kong Cantonese terms so children learn authentic vocabulary in their daily context.

Gap 2: AI for Preschool Learning

While LLMs and computer vision are growing in education, practical child-friendly use cases for preschoolers remain limited. The project translates AI capabilities into concrete learning loops.

Gap 3: Limited Digital Exposure Alternatives

The app experience is designed for short and focused interactions, helping families avoid unnecessary prolonged screen exposure while still enabling real-world vocabulary capture.

Project Objective

Develop an AI-powered Cantonese vocabulary learning platform centered on an app-based learning module, adaptive revision, personalized bedtime storytelling, and parent supervision analytics for sustained, measurable language growth.

Solution Architecture

The platform uses an app-based capture-to-learn flow and a centralized intelligence layer. It is composed of four core modules:

  • Learning Module
  • Revision Module
  • Bed-time Story Module
  • Supervision Module

These modules connect through shared learning records so that vocabulary captured during daily use can be reinforced through revision, bedtime narratives, and parent-facing analytics.

Methodology and User Journey

  • Onboarding and session continuity keep learning uninterrupted during capture and review.
  • After photo capture, on-device segmentation isolates the target object before recognition.
  • Server-side recognition returns Cantonese output in a stable child-friendly format.
  • The app displays a vocabulary card with one-tap audio playback for pronunciation practice.
  • Revision intelligence tracks mastery and supports association-based recommendations.
  • Daily bed-time stories are generated from learned vocabulary and delivered through TTS audio.
  • Supervision dashboards provide parents with streaks, word counts, and learning-time insights.

This workflow keeps the implementation practical while preserving a clear upgrade path to full backend intelligence and deployment in later phases.