International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 05 | May 2025
p-ISSN: 2395-0072
www.irjet.net
SKIN DISEASE DETECTION USING CNN MODEL Shrikant Bhaginath Bodkhe1, Dr. Sugandha Nandedkar2, Dr. Shaikh Shoaib3 1B. Tech in Computer Science and Engineering (Artificial Intelligence & Machine Learning), Deogiri Institute of Engineering and
Management Studies, Aurangabad, Maharashtra 2,3 Assistant Professor, Deogiri Institute of Engineering and Management Studies, Aurangabad, Maharashtra
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Skin diseases are one of the most common
and skin cancer, a WHO report says. In 2019, there were 4,859,267,654 new cases of skin and subcutaneous diseases (95% UI 4,680,693,440–5,060,498,767 cases) worldwide, of which bacterial and fungal skin diseases comprised the bulk of the new cases. A fourth make up 34% and fifth accounted for 23% of the total [1].
problems with human health around the world, affecting millions of indicisuals each year. Early and accurate diagnosis is extremely important in timely treatment and management. This project presents a deep learning-based system for automatic classification of skin diseases using convolutional neural networks (CNNs) that folds the MobilenetV2 as the base model. The data record consists of eight different classes of skin infections, including bacteria, fungi, parasitic and viral diseases. Images are subject to preprocessing steps such as resizing, normalization, and labeling coding to ensure consistency and optimal performance. The proposed framework reaches 97% accuracy and demonstrates its potential as an efficient and reliable means of supporting dermatologists in the diagnosis of skin diseases.
According to the report, "4,859,267,654 (95% uncertainty interval [UI], 4,680,693,440–5,060,498,767) new skin and subcutaneous disease cases that were identified, most were fungal (34.0%) and bacterial (23.0%) skin diseases, which accounted for 98,522 (95% UI 75,116–123,949) deaths. The burden of skin and subcutaneous diseases measured in DALYs was 42,883,695.48 (95% UI, 28,626,691.71– 63,438,210.22) in 2019, 5.26% of which were years of life lost, and 94.74% of which were years lived with disability" (Yakupu et al., 2023) [1].
Key Words: Skin Disease Detection, Convolutional Neural Network (CNN), MobileNet V2, Deep Learning, Image Classification, Medical Diagnosis, Artificial Intelligence, Accuracy 97%.
We have seen the growth of many technologies as medical science has advanced and many of these have enabled us to better diagnose those skin diseases. Despite this, many derms still depend on manual inspection to spot skin conditions. Though this archetypical approach shall always be the option of many practitioners, it isn’t flawless. It is a time consuming, menial and prone to human error, manual diagnosis. The outcome of which is that despite the complexity of certain dermatological conditions, it is not uncommon that different specialists come to very different conclusions about a diagnosis or treatment plan.
1.INTRODUCTION The skin is the largest organ of the human body, playing a crucial role as a protective barrier against bacteria, viruses, germs, and harmful ultraviolet (UV) radiation. Beyond protection, it performs several vital functions such as regulating body temperature, enabling the sensation of touch, and contributing to overall physical appearance. Healthy skin not only reflects good health but also significantly boosts self-confidence. Yet skin, while being important, is the subject of many diseases, most of which are increasingly common in today's world. There are some highly contagious skin conditions which have a very big public health impact. On many occasions, people miss their first dose or delay treatments in the early stages of cancer development because they are unaware or cannot afford treatment or they are too busy. If that is, negligence often leads to more serious complications, even life threatening ones, such as skin cancer. A lot of awareness needs to be raised concerning skin health, early diagnosis, and getting timely treatment with a view to avoiding such outcomes. In the case of skin care, I am happy to say that prioritizing it and acknowledging the importance of it as an important organ could lead to healthier lives and lessen skin disease burden on individual as well as the healthcare system.
A combination of clinical experience with unpredictable skin presentations may make the continued reliance on blind manual examination a reality. Skepticism towards fully automation has also been provoked by these factors. Though the potential benefits for integration of computer based diagnostic tools into dermatology may not be so obvious, the potential benefits are worthwhile. They also available in automated systems, much faster and more accurate in diagnosis with artificial intelligence (AI) and machine learning (ML) involved. With access to huge amounts of data, these models are able to identify delicate patterns that may be passed over for the human eye, helping to lower diagnostic mistakes and variety. Rather, the goal of AI/ML systems would be to act as a powerful decision support tool rather than replacing dermatologists. They might help clinicians make more consistent and precise diagnoses, ultimately helping patients. These technologies can become novel research
Each year, about 900 million people around the world have skin diseases – from simple conditions like acne to psoriasis
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 600