Alphabet-Sign-Language-Detection

Alphabet-Sign-Language-Detection is an image classification vision-language encoder model fine-tuned from google/siglip2-base-patch16-224 for a single-label classification task. It is designed to classify images into sign language alphabet categories using the SiglipForImageClassification architecture.

Classification Report:
              precision    recall  f1-score   support

           A     0.9995    1.0000    0.9998      4384
           B     1.0000    1.0000    1.0000      4441
           C     1.0000    1.0000    1.0000      3993
           D     1.0000    0.9998    0.9999      4940
           E     1.0000    1.0000    1.0000      4658
           F     1.0000    1.0000    1.0000      5750
           G     0.9992    0.9996    0.9994      4978
           H     1.0000    0.9979    0.9990      4807
           I     0.9992    1.0000    0.9996      4856
           J     1.0000    0.9996    0.9998      5227
           K     0.9972    1.0000    0.9986      5426
           L     1.0000    0.9998    0.9999      5089
           M     1.0000    0.9964    0.9982      3328
           N     0.9955    1.0000    0.9977      2635
           O     0.9998    1.0000    0.9999      4564
           P     1.0000    0.9993    0.9996      4100
           Q     1.0000    1.0000    1.0000      4187
           R     0.9998    0.9984    0.9991      5122
           S     0.9998    0.9998    0.9998      5147
           T     1.0000    1.0000    1.0000      4722
           U     0.9984    0.9998    0.9991      5041
           V     1.0000    0.9984    0.9992      5116
           W     0.9998    1.0000    0.9999      4926
           X     1.0000    0.9995    0.9998      4387
           Y     1.0000    1.0000    1.0000      5185
           Z     0.9996    1.0000    0.9998      4760

    accuracy                         0.9996    121769
   macro avg     0.9995    0.9996    0.9995    121769
weighted avg     0.9996    0.9996    0.9996    121769

The model categorizes images into the following 26 classes:

• Class 0: "A"
• Class 1: "B"
• Class 2: "C"
• Class 3: "D"
• Class 4: "E"
• Class 5: "F"
• Class 6: "G"
• Class 7: "H"
• Class 8: "I"
• Class 9: "J"
• Class 10: "K"
• Class 11: "L"
• Class 12: "M"
• Class 13: "N"
• Class 14: "O"
• Class 15: "P"
• Class 16: "Q"
• Class 17: "R"
• Class 18: "S"
• Class 19: "T"
• Class 20: "U"
• Class 21: "V"
• Class 22: "W"
• Class 23: "X"
• Class 24: "Y"
• Class 25: "Z"

Run with Transformers🤗

!pip install -q transformers torch pillow gradio

import gradio as gr
from transformers import AutoImageProcessor
from transformers import SiglipForImageClassification
from transformers.image_utils import load_image
from PIL import Image
import torch

# Load model and processor
model_name = "prithivMLmods/Alphabet-Sign-Language-Detection"
model = SiglipForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)

def sign_language_classification(image):
    """Predicts sign language alphabet category for an image."""
    image = Image.fromarray(image).convert("RGB")
    inputs = processor(images=image, return_tensors="pt")
    
    with torch.no_grad():
        outputs = model(**inputs)
        logits = outputs.logits
        probs = torch.nn.functional.softmax(logits, dim=1).squeeze().tolist()
    
    labels = {
        "0": "A", "1": "B", "2": "C", "3": "D", "4": "E", "5": "F", "6": "G", "7": "H", "8": "I", "9": "J",
        "10": "K", "11": "L", "12": "M", "13": "N", "14": "O", "15": "P", "16": "Q", "17": "R", "18": "S", "19": "T",
        "20": "U", "21": "V", "22": "W", "23": "X", "24": "Y", "25": "Z"
    }
    predictions = {labels[str(i)]: round(probs[i], 3) for i in range(len(probs))}
    
    return predictions

# Create Gradio interface
iface = gr.Interface(
    fn=sign_language_classification,
    inputs=gr.Image(type="numpy"),
    outputs=gr.Label(label="Prediction Scores"),
    title="Alphabet Sign Language Detection",
    description="Upload an image to classify it into one of the 26 sign language alphabet categories."
)

# Launch the app
if __name__ == "__main__":
    iface.launch()

Intended Use:

The Alphabet-Sign-Language-Detection model is designed for sign language image classification. It helps categorize images of hand signs into predefined alphabet categories. Potential use cases include:

• Sign Language Education: Assisting learners in recognizing and practicing sign language alphabets.
• Accessibility Enhancement: Supporting applications that improve communication for the hearing impaired.
• AI Research: Advancing computer vision models in sign language recognition.
• Gesture Recognition Systems: Enabling interactive applications with real-time sign language detection.