✅ Specialist detector (trained only on Nano) ✅ Unfrozen last attention layer for generator-specific features ✅ Test-Time Augmentation (TTA) enabled ✅ Heavy regularization (prevents overfitting on small dataset) ✅ H100-optimized training

Usage

python

from transformers import ViTForImageClassification, ViTImageProcessor
from PIL import Image
import torch

model = ViTForImageClassification.from_pretrained("ash12321/nano-banana-pro-specialist-h100")
processor = ViTImageProcessor.from_pretrained("google/vit-base-patch16-224")

image = Image.open("image.jpg")
inputs = processor(images=image, return_tensors="pt")

with torch.no_grad():
    outputs = model(**inputs)
    probs = torch.softmax(outputs.logits, dim=1)
    
if probs[0][1] > 0.5:
    print(f"AI-Generated (Nano): {probs[0][1]:.2%}")
else:
    print(f"Real: {probs[0][0]:.2%}")

Training Details

Training Time: 1.2 minutes
Best Epoch: 8
Device: H100 GPU
Unique data split (seed=123)

Limitations

⚠️ This is a specialist detector trained ONLY for Nano.

Does NOT detect:

Other AI image generators
General synthetic images

For comprehensive AI detection, use as part of an ensemble with other specialist detectors.