Devnagiri-Detection

Detecting Devnagiri(Hindi) characters, mother of more than 200 languages using Kaggle Dataset.

Character Type:

Metrics and training:

Final Classifier params:

Current Accuracy:

Current accuracy is subject to change over a longer period of training or using ensemble methods.

Saved Model

The trained model is 904.4 MB, I have saved the model in drive. Feel free to use it incase of lack of computational power. Link to the saved model : google drive

TO-DO:

1. Ensemble lower error models
2. Use Gans
3. Apply the same model for some other Dataset extensively.