Eco-AI Handwriting Digitizer: A Sustainable Handwritten Notes Summarization System Using On-Device Neural Models

Abstract

Handwritten notes remain an essential medium for learning, ideation, and documentation; however, the process of manually digitizing, organizing, and summarizing these notes is often time-consuming and environmentally unsustainable when done repeatedly on paper. This work proposes Eco-AI Handwrit- ing Digitizer, a lightweight on-device handwritten notes summa- rization system built using neural models optimized for energy efficiency and offline operation. The system captures handwritten input, performs preprocessing, extracts stroke-level and visual features, transcribes the handwriting into digital text, and gen- erates concise summaries without requiring cloud computation. By integrating edge-optimized models such as MobileViT and Tiny-BERT, the proposed system reduces latency, enhances data privacy, and minimizes carbon footprint by eliminating server- side inference. The experimental evaluation demonstrates high transcription accuracy and effective summarization quality while maintaining computational efficiency suitable for smartphones and low-power devices. The system aims to promote sustainability in digital learning environments by reducing paper usage and enabling eco-friendly digitization.