Singapore PE Teaching AI Assistant: Educational Innovation Integrating Computer Vision and Large Language Models

SG PE Syllabus Bot is an AI assistant tailored for Singaporean PE teachers. Integrating computer vision (Google MediaPipe) and large language model technologies, it addresses pain points in traditional physical education such as difficulty in motion assessment, cumbersome syllabus querying, and lack of personalized feedback. It enables real-time motion analysis and intelligent syllabus querying, providing innovative solutions for the digital transformation of physical education.

Traditional physical education faces three major challenges: difficulty in assessing motion standardization, low efficiency in accessing syllabi, and lack of personalized classroom feedback. The launch of Singapore's Ministry of Education 2024 PE Syllabus provided an opportunity for digital transformation. Against this backdrop, SG PE Syllabus Bot was born, combining AI technology with the actual needs of PE teaching to become an intelligent 'co-pilot' for teachers.

Core Functions

• Motion Skill Analysis: Through Google MediaPipe's real-time skeleton tracking, analyze students' Fundamental Movement Skills (FMS) frame by frame, identifying specific issues like "insufficient knee flexion".
• Teaching Policy Query: Digitize the 2024 PE Syllabus, supporting natural language queries to obtain information such as specific grade learning outcomes and sports teaching plans.

Technical Architecture

• Multi-Model Strategy: Support multiple large language models like Amazon Nova and Google Gemini 2.5 Flash, allowing selection of analysis engines as needed.
• Frontend Design: Built with React+TypeScript+Vite, with Tailwind CSS supporting light/dark themes, adapted to outdoor/gymnasium scenarios.
• Mobile-First: Optimized for tablets and mobile phones, enabling teachers to use it instantly on the teaching site.

1. Improved Teaching Efficiency: AI-assisted motion analysis provides an objective and detailed perspective, discovering details easily overlooked by teachers.
2. Enhanced Professional Support: Instant syllabus queries help teachers quickly access authoritative guidance, compensating for gaps in cross-sport professional knowledge.
3. Personalized Teaching Support: Identify students' specific issues through motion analysis, formulate targeted improvement plans, suitable for large-class teaching scenarios.

• Data Privacy: Student motion videos involve biometric data, so safe and compliant use must be ensured.
• Balance of Technology Dependence: AI should be an enhancement tool rather than replacing teachers' professional judgment, maintaining the humanistic care in education.
• Fairness: Avoid exacerbating educational resource imbalance due to technology access, ensuring all schools benefit equally.

Global Insights

The project demonstrates a replicable paradigm of "identifying teaching pain points + appropriate technology combination + focusing on user experience", providing reference for educational innovation in other regions.

Future Directions

• Expand support for more sports projects;
• Introduce student-side applications to provide personalized feedback;
• Build data analysis functions to track overall teaching effectiveness.

SG PE Syllabus Bot deeply understands the actual needs of physical education, integrates multiple AI technologies to solve pain points, demonstrates the broad possibilities of AI empowerment in the PE field (which traditionally receives less technical attention), and is a positive case of educational AI application worthy of attention from educators and technology developers.