Automated Detection and Diagnosis of Skin-Lesion using Transfer Learning based YOLOv7 Approach

Skin disease is a prevalent and growing health issue, its prompt and correct detection is crucial for treatment and therapy for that purpose. This paper, suggests a technique for identifying skin diseases that makes use of the YOLOv7 object identification algorithm’s capabilities. Visual inspection, which takes time and is prone to errors, has traditionally been used to detect skin diseases. As a result, automated skin disease detection utilizing computed methods has drawn a lot of importance in recent years. The goal of the suggested strategy is to offer a precise and efficient remedy for identifying skin diseases. For accurate recognition of various forms of skin diseases, the approach uses a self-labeled Dermnet data collection of skin diseases. In the proposed approach YOLOv7 models were trained via transfer learning, pre-trained templates, and big datasets. The experimental results demonstrate that the suggested YOLOv7-based method outperforms other methods in disease identification and achieves greater accuracy. As a result, dermatologists can quickly identify and treat skin diseases. This offers an accurate and efficient option for skin disease detection.