Rochard112/melanoma-model

Melanoma Detection Model

Model Information

• Architecture: EfficientNet-B0
• Format: ONNX
• Size: 16.59 MB
• Efficiency Score: 1.00
• Inference Time: 0.0715s

Optimization Targets

This model is optimized for Safe Scan validator incentives:

• F-Beta Score (β=2): 60% weight - prioritizes recall over precision
• Accuracy: 30% weight
• ROC-AUC: 10% weight
• Model Size: <50MB for full efficiency bonus

Usage

import onnxruntime as ort
import numpy as np

# Load model
session = ort.InferenceSession('melanoma_model.onnx')

# Preprocess image (512x512 RGB, normalized [0,1])
image = preprocess_image(your_image)  # Shape: (1, 3, 512, 512)

# Run inference
output = session.run(None, {'input': image})
prediction = output[0][0]  # Shape: (2,)

License

MIT License - Compatible with Safe Scan requirements

Competition

Safe Scan Subnet 76 - Melanoma Detection Competition