Hindi OCR - CRNN+CTC Model

This model performs Optical Character Recognition (OCR) on Hindi printed text using a CRNN+CTC architecture.

Model Details

  • Architecture: CRNN (Convolutional Recurrent Neural Network) with CTC (Connectionist Temporal Classification)
  • Input: 900×64 grayscale images
  • Output: Devanagari text with spaces
  • Training Data: 80,000 synthetic Hindi text line images

Training Progress

Check the metrics_epoch_*.txt files for detailed performance metrics.

Usage

See the training notebook for inference examples.

License

MIT License

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support