Ai Vs Human Image Detector

🚀 AI and Human Image Classification Model

A fine - tuned model trained on 60,000 AI - generated and 60,000 human images. It can effectively detect high - quality, state - of - the - art AI - generated images from popular models like Midjourney v6.1, Flux 1.1 Pro, Stable Diffusion 3.5, GPT - 4o, etc.

Detailed Training Code is available here: [blog/ai/fine - tuning - siglip2](https://exnrt.com/blog/ai/fine - tuning - siglip2/)

🚀 Quick Start

📦 Installation

pip install -q transformers torch Pillow accelerate

💻 Usage Examples

🔍 Basic Usage

import torch
from PIL import Image as PILImage
from transformers import AutoImageProcessor, SiglipForImageClassification

MODEL_IDENTIFIER = r"Ateeqq/ai-vs-human-image-detector"

# Device: Use GPU if available, otherwise CPU
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(f"Using device: {device}")

# Load Model and Processor
try:
    print(f"Loading processor from: {MODEL_IDENTIFIER}")
    processor = AutoImageProcessor.from_pretrained(MODEL_IDENTIFIER)

    print(f"Loading model from: {MODEL_IDENTIFIER}")
    model = SiglipForImageClassification.from_pretrained(MODEL_IDENTIFIER)
    model.to(device)
    model.eval()
    print("Model and processor loaded successfully.")

except Exception as e:
    print(f"Error loading model or processor: {e}")
    exit()

# Load and Preprocess the Image

IMAGE_PATH = r"/content/images.jpg" 
try:
    print(f"Loading image: {IMAGE_PATH}")
    image = PILImage.open(IMAGE_PATH).convert("RGB")
except FileNotFoundError:
    print(f"Error: Image file not found at {IMAGE_PATH}")
    exit()
except Exception as e:
    print(f"Error opening image: {e}")
    exit()

print("Preprocessing image...")
# Use the processor to prepare the image for the model
inputs = processor(images=image, return_tensors="pt").to(device)

# Perform Inference
print("Running inference...")
with torch.no_grad(): # Disable gradient calculations for inference
    outputs = model(**inputs)
    logits = outputs.logits

# Interpret the Results
# Get the index of the highest logit score -> this is the predicted class ID
predicted_class_idx = logits.argmax(-1).item()

# Use the model's config to map the ID back to the label string ('ai' or 'hum')
predicted_label = model.config.id2label[predicted_class_idx]

# Optional: Get probabilities using softmax
probabilities = torch.softmax(logits, dim=-1)
predicted_prob = probabilities[0, predicted_class_idx].item()

print("-" * 30)
print(f"Image: {IMAGE_PATH}")
print(f"Predicted Label: {predicted_label}")
print(f"Confidence Score: {predicted_prob:.4f}")
print("-" * 30)

# You can also print the scores for all classes:
print("Scores per class:")
for i, label in model.config.id2label.items():
    print(f"  - {label}: {probabilities[0, i].item():.4f}")

📊 Output

Using device: cpu
Model and processor loaded successfully.
Loading image: /content/images.jpg
Preprocessing image...
Running inference...
------------------------------
Image: /content/images.jpg
Predicted Label: ai
Confidence Score: 0.9996
------------------------------
Scores per class:
  - ai: 0.9996
  - hum: 0.0004

📚 Documentation

🔍 Evaluation Metrics

🏋️‍♂️ Train Metrics

• Epoch: 5.0
• Total FLOPs: 51,652,280,821 GF
• Train Loss: 0.0799
• Train Runtime: 2:39:49.46
• Train Samples/Sec: 69.053
• Train Steps/Sec: 4.316

📊 Evaluation Metrics (Fine - Tuned Model on Test Set)

• Epoch: 5.0
• Eval Accuracy: 0.9923
• Eval Loss: 0.0551
• Eval Runtime: 0:02:35.78
• Eval Samples/Sec: 212.533
• Eval Steps/Sec: 6.644

🔦 Prediction Metrics (on test set):

{
  "test_loss": 0.05508904904127121,
  "test_accuracy": 0.9923283699296264,
  "test_runtime": 167.1844,
  "test_samples_per_second": 198.039,
  "test_steps_per_second": 6.191
}

• Final Test Accuracy: 0.9923
• Final Test F1 Score (Macro): 0.9923
• Final Test F1 Score (Weighted): 0.9923

📄 License

This project is licensed under the Apache - 2.0 license.