DialecticalHTR is an HTR solution for Russian handwriting with diacritic characters (e.g. acutes, breves).
Made as a bachelor thesis at NaRFU by:
- Matvienko Daniil: Machine Learning
- Starikov Alexander: Dataset Building and implementation of client-server mobile app
The text of the bachelor thesis is available here: Google Drive
| Repository | Content |
| Dialect-HTR | Notebooks used for training TrOCR, YOLO and CRAFT models |
| RecognitionClient (not uploaded yet) | Android mobile application for recognition of dialectological handwritten materials |
| RecognitionBackend (not uploaded yet) | Server that uses the trained models for HTR |
