This repo is now maintained by IIT Delhi
Model checkpoints can be downloaded from here.
All models uploaded currently support Printed Text datasets. run.py can be modified to execute on multiple instances.
- English OCR
- Kannada OCR
- Tamil OCR
- Hindi OCR
- Assamese OCR
- Urdu OCR (For Urdu we have a seperate GitHub repo: UTRNet)
check_lmdb_to_jpg.py file can be used to convert mdb files to images. (Path to mdb folder consists of data.mdb and lock.mdb files)
Python version 3.9.0 is recommended
python run.py --model_path /path/to/model.ckpt --image_path /path/to/image.jpg