Digital note-taking is becoming popular due to its durability, editability, and easy organization. Derendering bridges the gap between digital and physical notes, converting handwriting into a digital form known as “digital ink.” This process creates a realistic representation of the user’s handwriting style, enhancing the durability and integration of notes with other digital content. A new approach to derendering allows the conversion of handwritten notes to digital ink without specialized equipment, offering a more natural writing experience. Challenges include limited supervised data and scalability to large images. The system workflow involves a vision-language model that combines recognition and derendering tasks to produce accurate results, comparable to human-generated digital ink data.
https://research.google/blog/a-return-to-hand-written-notes-by-learning-to-read-write/