Issuu

International Research Journal of Engineering and Technology (IRJET)

e-ISSN: 2395-0056

Volume: 12 Issue: 04 | Apr 2025

p-ISSN: 2395-0072

www.irjet.net

Deep Learning for Enhanced Dermatological Diagnostics 1st Dr.D.Durga Prasad∗, 2nd K.V.N.SSusanth∗, 3rd Sk.Ahmed Nazeer ∗, 4th K.Senapathi Reddy∗, 5th P.B.V Prasad ∗ Department of Computer Science and Engineering (CSE), PSCMR College of Engineering and Technology, Vijayawada, India ----------------------------------------------------------------------------***-----------------------------------------------------------------------Abstract—Early detection of dermatological conditions is crit- ical for treatment and management. But manual detection is time-consuming, subjective, and knowledge based. We present a system in this work that employs deep learning to identify and detect prevalent skin diseases using the YOLOv8 object detection model. We possess our database of twelve dermatological conditions, such as Acne, Chickenpox, Eczema, Monkeypox, Pimple, Psoriasis, Ringworm, Basal Cell Carcinoma, Melanoma, Tinea Versicolor, Vitiligo, and Warts. We employ the trained model of YOLOv8 coupled with an intuitive web interface developed using Flask, HTML, CSS, and JavaScript. Users can upload images of their skin from the interface, which displays the output image with disease detection annotations, symptoms, medications prescribed by a doctor, and precautions. It proves to be helpful for early detection, especially in remote or resource-limited environments, and can be of assistance to patients as well as medical professionals. We demonstrate that deep learning models like YOLOv8 can become a vital part of dermatological diagnoses by employing accessible, real-time solutions. KeyWords - Deep Learning, YOLOv8, Flask, Image Classification.

I. INTRODUCTION Dermatoses are among the most common diseases all over the world, and millions of individuals fall ill every year. Early diagnosis and treatment are extremely crucial to manage the diseases, as delayed or inappropriate diagnoses may result in complications and unjustified treatments. But computer diagnosis relies on specialist dermatologists and can contain a subjective interpretation of visual signals. As the pressure mounts to make quicker and more precise diagnoses, there is a growing need for automated technology that can support healthcare professionals, particularly in less affluent or rural areas where specialist access may be restricted.

detecting and classifying many diseases. Among these, object detection architectures such as YOLO (You Only Look Once) have shown outstanding ability to detect objects in images in real time. The advanced YOLOv8 model is fast and accurate and, thus, an appropriate tool to detect skin disease. In this paper, we introduce the use of YOLOv8 for object detection and classification of twelve common skin diseases such as Acne, Chickenpox, Eczema, Monkeypox, Pimple, Psoriasis, Ringworm, Basal Cell Carcinoma, Melanoma, Tinea Versicolor, Vitiligo, and Warts. We train the model on a custom dataset and introduce a simple webbased UI implemented using Flask, HTML, CSS, and JavaScript. This interface allows for the uploading of skin images, which are then used by the model to identify whether or not a certain disease exists. The output is the identified disease, along with related information such as symptoms, drugs, and precautions. This study aims to offer a practical, easy-to-use tool for both medical professionals and patients to identify the disease at an early stage and enhance the overall quality of diagnostics. With real-time object detection and deep learning, we expect the efficiency of dermatological diagnostics to be elevated, particularly in environments where professional medical expertise is not easily accessible.

II. LITERATURE SURVEY In the few years, deep learning algorithms, especially convolutional neural networks (CNNs), have drastically changed the nature of medical image analysis. More so in the case of dermatology, in which deep learning algorithms have had a lot of potential for applying machine learning on skin disease detection. Several works have been concentrating on using such techniques for detecting, segmenting, and classifying skin disorders from images.

Deep learning methods, especially convolutional neural networks (CNNs), have been highly promising in the area of medical image analysis, with greater accuracy and speed in

The paper titled ”Machine Learning and Deep Learning Integration for Skin Diseases Prediction”[1] by SamirKumar Bandyopadhyay, Payal Bose, Amiya Bhaumik, and Sandeep Poddar explores the integration of Google Net Inception V3

ISO 9001:2008 Certified Journal

Impact Factor value: 8.315

Page 1971