Skin Cancer Classifier — ViT-Large
Fine-tuned google/vit-large-patch16-224 on the HAM10000 dataset (10,000+ dermoscopy images, 7 classes). Achieved 92.74% accuracy and 92.60% weighted F1 on the validation set.
Upload a dermoscopy skin image to classify it into one of 7 categories.
⚠️ This is a portfolio/learning project — not a medical tool. Do not use for real diagnosis.
🛡️ Non-skin images are automatically rejected by a CLIP-based validation gate.