Natural Data-augmentation for Skin Lesions (ISIC-2017 Challenge Dataset)
Description
Data augmentation techniques may mitigate many limitations of datasets, such as imbalanced data among the classes of skin lesions and heterogeneous sources of data, by adding augmented samples with different image transformations, such as rotation, random crop, horizontal and vertical flip, translation, shear, color jitter, and colorspace. It is proven in many studies that data augmentation improved the diagnosis of skin cancer. In the HAM10000 dataset, the skin lesion images were captured at different magni fications or angles or with different cameras, a process known as natural data augmentation. We used a deep learning architecture called Faster R-CNN to develop the algorithm to generate augmented copies similar to the natural data-augmentation for ISIC-2017 challenge dataset.