Khalsa-Phulwari/gurumkhi-recognizer

Gurmukhi Character Recognition Model

This model is a Vision Transformer (ViT) fine-tuned for recognizing Gurmukhi characters from images.

Model Details

• Model Type: Vision Transformer (ViT)
• Language: Punjabi (Gurmukhi script)
• Task: Image Classification / Character Recognition
• Input Size: 224x224 pixels
• Number of Classes: 41 Gurmukhi characters

Supported Characters

The model can recognize 41 Gurmukhi characters:

• Basic consonants: ਕ, ਖ, ਗ, ਘ, ਙ, ਚ, ਛ, ਜ, ਝ, ਞ, ਟ, ਠ, ਡ, ਢ, ਣ, ਤ, ਥ, ਦ, ਧ, ਨ, ਪ, ਫ, ਬ, ਭ, ਮ, ਯ, ਰ, ਲ, ਵ, ਸ, ਹ
• Vowels: ੳ, ਅ, ੲ
• Additional characters: ੜ, ਸ਼, ਖ਼, ਗ਼, ਜ਼, ਫ਼, ਲ਼

Usage

from transformers import ViTForImageClassification, ViTFeatureExtractor
from PIL import Image

# Load model and feature extractor
model = ViTForImageClassification.from_pretrained("Khalsa-Phulwari/gurumkhi-recognizer")
feature_extractor = ViTFeatureExtractor.from_pretrained("Khalsa-Phulwari/gurumkhi-recognizer")

# Process image
image = Image.open("path/to/gurmukhi_character.png").convert("RGB")
inputs = feature_extractor(images=image, return_tensors="pt")

# Get prediction
outputs = model(**inputs)
predicted_class = outputs.logits.argmax(-1).item()

Training Details

• Framework: PyTorch + Transformers
• Base Model: Vision Transformer (ViT)
• Training Data: Custom Gurmukhi character dataset
• Image Preprocessing: Resize to 224x224, normalize

Model Performance

• Accuracy: ~95% (update with your actual metrics)
• Inference Time: ~50ms per image (CPU)
• Model Size: ~327MB

Limitations

• Works best with clear, well-cropped character images
• Performance may vary with handwritten vs. printed text
• Optimized for single character recognition

Citation

If you use this model, please cite:

@misc{gurmukhi-vit-recognizer,
  title={Gurmukhi Character Recognition using Vision Transformer},
  author={Your Name},
  year={2024},
  url={https://huggingface.co/Khalsa-Phulwari/gurumkhi-recognizer}
}