Spaces:

TruthLens
/

fake-real

Runtime error

App Files Files Community

TruthLens commited on Feb 24

Commit

723860c

verified ·

1 Parent(s): d59602f

Update app.py

Browse files

Files changed (1) hide show

app.py +24 -70

app.py CHANGED Viewed

@@ -2,8 +2,7 @@ import os
 from flask import Flask, request, render_template_string
 from PIL import Image
 import torch
-from torchvision import transforms
-from transformers import pipeline, CLIPProcessor, CLIPModel
 app = Flask(__name__)
@@ -11,61 +10,28 @@ app = Flask(__name__)
 upload_folder = os.path.join('static', 'uploads')
 os.makedirs(upload_folder, exist_ok=True)
-# Fake News Detection Models
-news_models = {
-    "mrm8488": pipeline("text-classification", model="mrm8488/bert-tiny-finetuned-fake-news-detection"),
-    "google-electra": pipeline("text-classification", model="google/electra-base-discriminator"),
-    "bert-base": pipeline("text-classification", model="bert-base-uncased")
-}
-# Image Model for AI vs. Human Detection
 clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
 clip_processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
-# Image transformation pipeline
-transform = transforms.Compose([
-    transforms.Resize((224, 224)),
-    transforms.ToTensor(),
-    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
-])
-# HTML Template
 HTML_TEMPLATE = """
 <!DOCTYPE html>
 <html lang="en">
 <head>
     <meta charset="UTF-8">
-    <title>AI & News Detection</title>
     <style>
         body { font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif; background-color: #f5f5f5; padding: 20px; }
         .container { background: white; padding: 30px; border-radius: 12px; max-width: 850px; margin: auto; box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1); }
-        textarea, select { width: 100%; padding: 12px; margin-top: 10px; border-radius: 8px; border: 1px solid #ccc; }
-        button { background-color: #4CAF50; color: white; border: none; padding: 12px 20px; border-radius: 8px; cursor: pointer; font-size: 16px; margin-top: 10px; }
         button:hover { background-color: #45a049; }
         .result { background: #e7f3fe; padding: 15px; border-radius: 10px; margin-top: 20px; }
     </style>
 </head>
 <body>
     <div class="container">
-        <h1>📰 Fake News Detection</h1>
-        <form method="POST" action="/detect">
-            <textarea name="text" placeholder="Enter news text..." required></textarea>
-            <label for="model">Select Fake News Model:</label>
-            <select name="model" required>
-                <option value="mrm8488">MRM8488 (BERT-Tiny)</option>
-                <option value="google-electra">Google Electra (Base Discriminator)</option>
-                <option value="bert-base">BERT-Base Uncased</option>
-            </select>
-            <button type="submit">Detect News Authenticity</button>
-        </form>
-        {% if news_prediction %}
-        <div class="result">
-            <h2>🧠 News Detection Result:</h2>
-            <p>{{ news_prediction }}</p>
-        </div>
-        {% endif %}
         <h1>🖼️ AI vs. Human Image Detection</h1>
         <form method="POST" action="/detect_image" enctype="multipart/form-data">
             <input type="file" name="image" required>
@@ -76,7 +42,7 @@ HTML_TEMPLATE = """
         <div class="result">
             <h2>📷 Image Detection Result:</h2>
             <p>{{ image_prediction }}</p>
-            <p><strong>Explanation:</strong> The model estimates image complexity by analyzing feature variability. Higher complexity typically indicates human-created content, while smoother, less varied features suggest AI generation.</p>
         </div>
         {% endif %}
     </div>
@@ -88,23 +54,6 @@ HTML_TEMPLATE = """
 def home():
     return render_template_string(HTML_TEMPLATE)
-@app.route("/detect", methods=["POST"])
-def detect():
-    text = request.form.get("text")
-    model_key = request.form.get("model")
-    if not text or model_key not in news_models:
-        return render_template_string(HTML_TEMPLATE, news_prediction="Invalid input or model selection.")
-    result = news_models[model_key](text)[0]
-    label = "REAL" if result['label'].lower() in ["real", "label_1", "neutral"] else "FAKE"
-    confidence = result['score'] * 100
-    return render_template_string(
-        HTML_TEMPLATE,
-        news_prediction=f"News is {label} (Confidence: {confidence:.2f}%)"
-    )
 @app.route("/detect_image", methods=["POST"])
 def detect_image():
     if "image" not in request.files:
@@ -112,22 +61,27 @@ def detect_image():
     file = request.files["image"]
     img = Image.open(file).convert("RGB")
-    inputs = clip_processor(images=img, return_tensors="pt")
     with torch.no_grad():
-        image_features = clip_model.get_image_features(**inputs)
-        feature_variance = torch.var(image_features).item()
-        prediction = "Human-Created" if feature_variance > 0.05 else "AI-Generated"
-    explanation = (
-        f"Prediction: {prediction} (Feature Variance: {feature_variance:.4f}). "
-        "Higher variance indicates complex, diverse features typical of human-created images, while lower variance suggests smoother, AI-generated patterns."
-    )
-    return render_template_string(
-        HTML_TEMPLATE,
-        image_prediction=explanation
     )
 if __name__ == "__main__":
-    app.run(host="0.0.0.0", port=7860)

 from flask import Flask, request, render_template_string
 from PIL import Image
 import torch
+from transformers import CLIPProcessor, CLIPModel
 app = Flask(__name__)
 upload_folder = os.path.join('static', 'uploads')
 os.makedirs(upload_folder, exist_ok=True)
+# Load CLIP model and processor
 clip_model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
 clip_processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+# HTML Template with Model Selection and Explanations
 HTML_TEMPLATE = """
 <!DOCTYPE html>
 <html lang="en">
 <head>
     <meta charset="UTF-8">
+    <title>AI vs. Human Image Detection</title>
     <style>
         body { font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif; background-color: #f5f5f5; padding: 20px; }
         .container { background: white; padding: 30px; border-radius: 12px; max-width: 850px; margin: auto; box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1); }
+        input[type="file"], button { width: 100%; padding: 12px; margin-top: 10px; border-radius: 8px; border: 1px solid #ccc; }
+        button { background-color: #4CAF50; color: white; border: none; font-size: 16px; cursor: pointer; }
         button:hover { background-color: #45a049; }
         .result { background: #e7f3fe; padding: 15px; border-radius: 10px; margin-top: 20px; }
     </style>
 </head>
 <body>
     <div class="container">
         <h1>🖼️ AI vs. Human Image Detection</h1>
         <form method="POST" action="/detect_image" enctype="multipart/form-data">
             <input type="file" name="image" required>
         <div class="result">
             <h2>📷 Image Detection Result:</h2>
             <p>{{ image_prediction }}</p>
+            <p><strong>Explanation:</strong> The model compares the image to text prompts representing AI-generated and human-created images. The prediction is based on which prompt the image is more similar to.</p>
         </div>
         {% endif %}
     </div>
 def home():
     return render_template_string(HTML_TEMPLATE)
 @app.route("/detect_image", methods=["POST"])
 def detect_image():
     if "image" not in request.files:
     file = request.files["image"]
     img = Image.open(file).convert("RGB")
+    # Text prompts for comparison
+    prompts = ["AI-generated image", "Human-created image"]
+    # Process image and text inputs
+    inputs = clip_processor(text=prompts, images=img, return_tensors="pt", padding=True)
     with torch.no_grad():
+        outputs = clip_model(**inputs)
+        logits_per_image = outputs.logits_per_image.squeeze()
+        probs = logits_per_image.softmax(dim=0)
+    ai_score, human_score = probs.tolist()
+    prediction = "AI-Generated" if ai_score > human_score else "Human-Created"
+    result_text = (
+        f"Prediction: {prediction} <br>"
+        f"AI Similarity: {ai_score * 100:.2f}% | Human Similarity: {human_score * 100:.2f}%"
     )
+    return render_template_string(HTML_TEMPLATE, image_prediction=result_text)
 if __name__ == "__main__":
+    app.run(host="0.0.0.0", port=7860)  # Suitable for public launch