app.py

import streamlit as st
from transformers import ViTImageProcessor, AutoModelForImageClassification
from PIL import Image
import requests
from io import BytesIO
import json

# Load the model and processor
processor = ViTImageProcessor.from_pretrained('AdamCodd/vit-base-nsfw-detector')
model = AutoModelForImageClassification.from_pretrained('AdamCodd/vit-base-nsfw-detector')

# Define prediction function
def predict_image(image):
    try:
        # Process the image and make prediction
        inputs = processor(images=image, return_tensors="pt")
        outputs = model(**inputs)
        logits = outputs.logits

        # Get predicted class
        predicted_class_idx = logits.argmax(-1).item()
        predicted_label = model.config.id2label[predicted_class_idx]

        return predicted_label
    except Exception as e:
        return str(e)

# Streamlit app for UI and API endpoint
st.title("NSFW Image Classifier")

# URL input for UI
image_url_ui = st.text_input("Enter Image URL", placeholder="Enter image URL here")

# API endpoint for classification (POST request)
@st.experimental_singleton  # Ensure a single instance for performance
def api_endpoint():
    if request.method == 'POST':
        data = request.json
        if 'image_url' in data:
            try:
                image_url = data['image_url']
                # Load image from URL
                response = requests.get(image_url)
                image = Image.open(BytesIO(response.content))

                # Predict and return result as JSON
                prediction = predict_image(image)
                return json.dumps({'predicted_class': prediction})
            except Exception as e:
                return json.dumps({'error': str(e)}), 500  # Internal Server Error
        else:
            return json.dumps({'error': 'Missing "image_url" in request body'}), 400  # Bad Request
    else:
        return json.dumps({'error': 'Only POST requests are allowed'}), 405  # Method Not Allowed

st.experimental_next_router(api_endpoint)  # Register the API endpoint

if image_url_ui:
    try:
        # Load image from UI input (if URL is provided)
        response = requests.get(image_url_ui)
        image = Image.open(BytesIO(response.content))
        st.image(image, caption='Image from URL', use_column_width=True)
        st.write("")
        st.write("Classifying...")

        # Predict and display result (for UI)
        prediction = predict_image(image)
        st.write(f"Predicted Class: {prediction}")
    except Exception as e:
        st.write(f"Error: {e}")

# Display API endpoint information
space_url = st.session_state.get('huggingface_space_url')  # Assuming it's available
if space_url:
    api_endpoint_url = f"{space_url}/api/classify"  # Construct the URL based on Space URL
    st.write(f"You can also use this API endpoint to classify images:")
    st.write(f"```curl")
    st.write(f"curl -X POST -H 'Content-Type: application/json' -d '{{ \"image_url\": \"https://example.jpg\" }}' {api_endpoint_url}")
    st.write(f"```")
    st.write(f"This will return the predicted class in JSON format.")