Skin Lesion Classifier (ResNet18, HAM10000, 7 classes)
Trained on HAM10000. Educational project — not diagnostic-grade, not a medical device.
Classes: {'actinic_keratoses': 0, 'basal_cell_carcinoma': 1, 'benign_keratosis-like_lesions': 2, 'dermatofibroma': 3, 'melanocytic_Nevi': 4, 'melanoma': 5, 'vascular_lesions': 6}
Load
import torch, torchvision.models as models
from huggingface_hub import hf_hub_download
model = models.resnet18(weights=None)
model.fc = torch.nn.Linear(model.fc.in_features, 7)
weights_path = hf_hub_download(repo_id="Farhan2000/skin-lesion-classifier-resnet18", filename="model.pth")
model.load_state_dict(torch.load(weights_path, map_location="cpu"))
model.eval()
Validation accuracy: 0.8937