High_Res-vs-Low_Res

High_Res-vs-Low_Res 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 assess the resolution quality of images using the SiglipForImageClassification architecture.

The model categorizes images into two classes:

• Class 0: "High Resolution Image" – indicating that the image has a high resolution and appears sharp and detailed.
• Class 1: "Low Resolution Image" – indicating that the image has a low resolution and may appear pixelated or blurry.

    Classification Report:
                           precision    recall  f1-score   support
    
    high resolution image     0.5697    0.5407    0.5548      1254
     low resolution image     0.8495    0.8639    0.8566      3762
    
                 accuracy                         0.7831      5016
                macro avg     0.7096    0.7023    0.7057      5016
             weighted avg     0.7795    0.7831    0.7812      5016

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/High_Res-vs-Low_Res"
model = SiglipForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)

def resolution_classification(image):
    """Predicts image resolution classification."""
    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": "High Resolution Image", "1": "Low Resolution Image"}
    predictions = {labels[str(i)]: round(probs[i], 3) for i in range(len(probs))}
    
    return predictions

# Create Gradio interface
iface = gr.Interface(
    fn=resolution_classification,
    inputs=gr.Image(type="numpy"),
    outputs=gr.Label(label="Prediction Scores"),
    title="Image Resolution Classification",
    description="Upload an image to classify its resolution quality."
)

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

Intended Use:

The High_Res-vs-Low_Res model is designed to evaluate the resolution quality of images. It helps distinguish between high-resolution and low-resolution images. Potential use cases include:

• Image Quality Assessment: Identifying whether an image meets high-resolution standards or suffers from low-quality artifacts.
• Content Moderation: Assisting platforms in filtering low-resolution images for better user experience.
• Forensic Analysis: Supporting researchers and analysts in determining the clarity of images used in various applications.
• Image Processing Pipelines: Helping developers optimize image enhancement algorithms by assessing resolution quality.