Japanese Kana Recognition
Project Scope:
This project applies neural network machine learning using CNN to correctly identify Japanese characters. The three writing systems of Japan (hiragana, katakana, kanji) are represented and modeled individually.
This project aims to tackle the issue of accurately transcribing handwritten characters and to be identified by image modeling and NLP. In a linguistic and archival sense, this would do well to preserve ancient writings and transcribe important texts much more quickly.
Also connected to linguistics, can this model serve as a base for an educational handwriting recognition API (ie. used in a dictionary app for searching by radical, mora, and character).