Skin Cancer Classification with Deep Learning
Francisco Luis de Avelar Cardoso
Published
0
An applied computer vision and multimodal learning project for automated skin lesion classification.
This project compares:
A CV-only ResNet50 model trained on dermatoscopic images
A multimodal model combining images with clinical metadata (age, sex, localization)
Includes a Streamlit dashboard for EDA, model comparison, confusion matrices, per-class F1 scores, and live predictions, highlighting the practical trade-offs between visual and tabular information in medical AI.