Melanoma Image Augmentation and Classification

• Github URL: Project Link

Trained a Cycle Generative Adversarial Network to perform image to image translation between benign and malign skin lesion images.

Performed data augmentation by generating synthetic malign samples and balanced the highly imbalanced SIIM‑ISIC Melanoma dataset.

Fine‑tuned pre‑trained EfficientNet weights for the binary classification task and obtained ROC‑AUC of 0.89 on the test set.