An Unified Integrated Learning Approach for Multi-Class Skin Disease Detection

Skin diseases or cancers are becoming more common in human daily life. These are also called skin abnormalities, and there are many causes, such as pollution, vitamin deficiencies, and heredity. Many researchers have conducted numerous studies to detect skin abnormalities at early stages, but they have not identified the exact causes. In this paper, we present a unified integrated learning approach (UILA) to classify multiple skin diseases and cancers. The proposed UILA introduced the new USkinLearn-Classify model, which learns the unique patterns in input skin images. In this context, the DermoNet is a pre-trained deep learning (DL) model that detects multiple skin abnormalities using various layers. In the preprocessing step, researchers use an advanced preprocessing technique to filter noise and improve image quality. The SkinNetV1 segmentation technique segments skin lesions from input images. The SkinNet is an integrated approach that combines advanced convolutional blocks, dense feature connections, and attention-driven mechanisms. These functionalities help the proposed USkinLearn-Classify detect and classify multi-class skin diseases. The USkinLearn contains multiple layers that learn and identify every lesion with a high classification rate. This paper uses the benchmark dataset, and the proposed approach achieves high performance.