SCHOOL OF ENGINEERING AND TECHNOLOGY

A PROJECT REPORT
On
"AI ASSISTED SKIN DISEASE DETECTION SYSTEM"

Submitted in partial fulfillment of the requirements for the Course

SOFTWARE ENGINEERING FOR AI (4AIML2061)

Bachelor of Technology
In
Computer Science and Engineering (AI & ML)
SoET, CMR University, Bangalore

Submitted by:

Milind A Devoor (22BBTCA095)

Mohammed Afnan Sayeed (22BBTCA096)
Mohammed Anas (22BBTCA097)
Niranjan.S (22BBTCA108)

Under the Supervision:

Dr. ELANGO.S
Assistant Professor
Department of CSE-AIML

Department of Computer Science and Engineering (AI & ML)

Off Hennur - Bagalur Main Road,
Near Kempegowda International Airport, Chagalahatti,
Bangalore, Karnataka-562149
2024-2025
 SCHOOL OF ENGINEERING AND TECHNOLOGY
Chagalahatti, Bengaluru, Karnataka- 562149

Department of Computer Science and Engineering (AI & ML)

CERTIFICATE
This is to certify that the project report entitled “AI ASSISTED SKIN DISEASE

DETECTION SYSTEM”, is a record of work successfully carried out by Milind A Devoor

(22BBTCA095), Mohammed Afnan Sayeed (22BBTCA096), Mohammed Anas (22BBTCA097),

Niranjan.S (22BBTCA108) in partial fulfilment of the requirement for the course SOFTWARE

ENGINEERING FOR AI (4AIML2061) of Bachelor of Technology in Computer Science and

Engineering (AI & ML), SoET, CMR University, Bangalore during the academic year 2024-25, under

the supervision and guidance of Dr. ELANGO.S, Assistant Professor, CSE(AI & ML), SoET, CMR

University.

Signature Signature
Dr. Elango.S, Dr. Manjunath C.R,
Assistant Professor, Professor and Head,
Dept of CSE(AI & ML) Dept of CSE(AI & ML)

PAGE \* MERGEFORMAT i
 TABLE OF CONTENT

Chapter Title Page No

No

ABSTRACT 3

1 INTRODUCTION 4
1.1 Background and Context 4
1.2 Objectives of the study 5

2 LITERATURE REVIEW 6
2.1 Introduction 6
2.2 Historical Evolution and Statistical Foundations 6
2.3 Machine Learning Models for skin disease detection 6
2.4 Challenges in Skin disease Detection 6
2.5 Role of NLP in Symptom-Based Detection 7
2.6 Future Directions 7

3 ABOUT DESIGN TOOLS USED 8

4 SYSTEM DESIGN 10

5 SYSTEM ANALYSIS 13

6 IMPLEMENTATION PLAN 16

7 CONCLUSION 19

8 REFERENCES 20

PAGE \* MERGEFORMAT iv
 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

ABSTRACT
Skin diseases affect a large percentage of the global population, yet early and accurate diagnosis
remains a challenge due to limited access to dermatologists and specialist care. The AI Assisted
Skin Disease Detection System aims to bridge this gap by providing an intelligent platform that can
assist in preliminary disease identification. Using machine learning techniques, the system analyzes
patient symptoms and/or skin images to predict probable skin conditions, supporting faster medical
intervention.
The project leverages advanced algorithms for classification, feature extraction, and predictive
analysis. It integrates image processing, text-based symptom interpretation, and real-time
recommendation features to deliver a comprehensive diagnostic aid. This system not only supports
patients in underserved areas but also enhances the efficiency of clinical workflows.
By reducing diagnosis time and improving preliminary detection accuracy, the AI Assisted Skin
Disease Detection System plays an essential role in improving healthcare accessibility, minimizing
misdiagnosis, and promoting early treatment for various dermatological conditions.

Keywords:

 Artificial Intelligence (AI)

 Machine Learning
 Real-time Data
 Mobile Application

Dept of CSE-AIML, SoET, CMRU 2024-25 3

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

CHAPTER – 1
INTRODUCTION

1.1 Background and Context

Skin diseases such as eczema, psoriasis, dermatitis, acne, and fungal infections are among the most
common health problems globally. Accurate and early diagnosis plays a crucial role in effective
treatment and management. However, the shortage of dermatologists, especially in rural and
underserved regions, often leads to delays in diagnosis, worsening patient outcomes.

The advancement of Artificial Intelligence (AI) in healthcare provides an opportunity to automate

and assist in the detection of skin diseases. AI technologies, especially Machine Learning (ML) and
Deep Learning (DL), can analyze patient-provided data—either through images or symptom
descriptions—and offer preliminary diagnostic support. By integrating AI tools into dermatological
services, healthcare systems can increase accessibility, speed up diagnosis, and provide better patient
care.

The AI Assisted Skin Disease Detection System is developed with this vision. It combines image
analysis and symptom-based assessment models to detect and suggest possible skin conditions
accurately, offering a scalable solution that fits into the broader context of smart healthcare and
telemedicine.

1.2 Problem Definition

Traditional diagnosis methods heavily rely on expert examination and laboratory tests, which may
not be accessible or affordable for all. Moreover, manual diagnosis can be subjective and vary
depending on the expertise of the healthcare provider. This situation highlights the urgent need for an
automated, intelligent system that can:

 Assist healthcare professionals with initial disease prediction.

 Help patients get quick preliminary assessments.
 Improve diagnostic consistency and objectivity.

Thus, the core problem is the lack of accessible, intelligent, and reliable skin disease detection tools
for mass use.

1.3 Objectives

Dept of CSE-AIML, SoET, CMRU 2024-25 4

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

The primary aim of this project is to build an intelligent and accessible AI-powered system that can
assist users in identifying potential skin diseases based on either uploaded skin images or described
symptoms. This tool will act as a supportive diagnostic platform that can enhance healthcare access
and early intervention, especially in regions with limited dermatological services. The following are
the detailed objectives of the system:

 To develop an intelligent skin disease detection system that leverages the power of machine
learning and image processing techniques to identify various dermatological conditions. This
includes building and training classification models using image datasets of skin infections,
acne, rashes, and other conditions, ensuring that the model learns to detect patterns and
features associated with each disease accurately.

 To create a solution that is low-cost, efficient, and scalable, ensuring it can be deployed
across various platforms, including mobile applications, web portals, and remote clinics. By
doing so, the system can extend its reach to rural and underserved areas where access to
dermatologists is limited or non-existent.

 To facilitate early and accessible diagnosis by enabling users to either upload a photo of their
skin condition or provide a description of symptoms through a text-based interface. This
dual-input functionality allows for more flexibility, making the system useful in both image-
rich and text-based environments (such as chatbots or mobile apps).

 To support healthcare professionals and improve decision-making by offering preliminary

diagnostic insights based on AI analysis. Although not intended to replace clinical judgment,
the system can serve as a second opinion or triage tool to prioritize patients and enhance
diagnosis efficiency.

 To design a user-friendly interface that offers real-time feedback to the user after analyzing
the inputs. The system will generate predictions on possible skin conditions along with
confidence scores and appropriate recommendations, such as visiting a dermatologist or using
over-the-counter treatments. The interface will be designed to be intuitive even for users with
minimal technical knowledge.

 To ensure that the system can adapt and improve over time by incorporating user feedback,
new data, and updated medical knowledge. This includes integrating additional disease
categories, refining prediction accuracy, and expanding multilingual support to reach diverse
user groups.

Through these objectives, the project aims to contribute to the broader goal of smart healthcare
systems by combining AI, medical informatics, and human-centered design.

CHAPTER – 2

Dept of CSE-AIML, SoET, CMRU 2024-25 5

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

LITERATURE SURVEY

2.1. Introduction

The detection and diagnosis of skin diseases have traditionally depended on physical examination
and laboratory analysis. However, with the rapid advancement of Artificial Intelligence (AI) and
Machine Learning (ML) technologies, automated systems capable of assisting in medical diagnoses
have emerged. These systems offer significant benefits in terms of speed, scalability, and cost-
effectiveness, especially for dermatology, where visual inspection is critical.

2.2. Historical Evolution and Statistical Foundations

Early AI efforts in dermatology focused primarily on rule-based expert systems that depended on
predefined symptom patterns. As machine learning algorithms matured, Support Vector Machines
(SVM) and Random Forest classifiers became popular for disease prediction using structured
datasets.
With the rise of Deep Learning, especially Convolutional Neural Networks (CNNs), systems
began to match and even surpass dermatologist-level accuracy in tasks like skin lesion classification.
Esteva et al. (2017) demonstrated that a deep CNN trained on skin lesion images could achieve
performance comparable to certified dermatologists in melanoma detection.

2.3. Machine Learning Models for Skin Disease Detection

Several studies have shown the effectiveness of machine learning in skin disease prediction:

 CNN-based models: Used primarily for image-based diagnosis; capable of learning complex
skin pattern features automatically.
 Decision Tree and Random Forest classifiers: Effective for structured data like patient
symptoms.
 Ensemble methods: Combining multiple models to improve prediction accuracy.
 Transfer learning: Using pre-trained models (like ResNet, VGG) fine-tuned on dermatology
datasets to improve performance on limited medical data.

2.4. Challenges in Skin Disease Detection

Despite advancements, challenges still exist:

 Variability in skin tone and image quality can affect model accuracy.
 Scarcity of labeled medical datasets.
 Need for high precision in distinguishing between visually similar conditions (e.g., psoriasis
vs. eczema).
 Ethical concerns regarding patient data privacy and AI-driven diagnosis bias.

2.5. Role of NLP in Symptom-Based Detection

Dept of CSE-AIML, SoET, CMRU 2024-25 6

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

Beyond image processing, Natural Language Processing (NLP) techniques are being used to
interpret patient-reported symptoms. NLP models extract meaningful patterns from textual
descriptions and predict possible conditions, making it easier for non-visual-based platforms (like
mobile apps or chatbots) to assist in diagnosis.

2.6. Future Directions

Ongoing research aims to:

 Improve robustness of models across diverse demographics.

 Combine multimodal inputs (text + images) for higher accuracy.

 Deploy AI systems on mobile devices for real-time use.

 Incorporate explainable AI (XAI) to make model decisions interpretable for healthcare

providers.

CHAPTER - 3

ABOUT DESIGN TOOLS USED

Dept of CSE-AIML, SoET, CMRU 2024-25 7

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

The successful development of the AI Assisted Skin Disease Detection System relies on a
combination of software tools, machine learning libraries, and development environments. Each tool
and technology was selected based on its relevance to image processing, machine learning, and
system deployment.

1. Python

Python was chosen as the primary programming language due to its simplicity, versatility, and rich
ecosystem for machine learning and AI development. It provides powerful libraries for both image
analysis and natural language processing.

2. Scikit-learn

Scikit-learn is an essential machine learning library offering simple and efficient tools for data
mining, analysis, and model building. It was used for:
 Training machine learning classifiers like Logistic Regression and Decision Trees.
 Performing data preprocessing and feature selection.
 Evaluating model performance using metrics such as accuracy, precision, and recall.

3. TensorFlow / Keras

For deep learning tasks, especially image-based classification, TensorFlow and Keras libraries were
used. They supported:
 Building Convolutional Neural Networks (CNNs) for analyzing skin images.
 Implementing transfer learning models like ResNet and MobileNet.
 Accelerating model training using GPU capabilities.

4. OpenCV

OpenCV (Open Source Computer Vision Library) is a powerful library for image processing tasks. It
was utilized for:
 Reading and preprocessing skin images.
 Resizing, normalizing, and augmenting images for training.
 Detecting skin regions and enhancing image quality where necessary.

5. Pandas and NumPy

Pandas and NumPy were critical for:

Dept of CSE-AIML, SoET, CMRU 2024-25 8

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

 Handling large structured datasets containing symptom descriptions and disease labels.
 Performing efficient data manipulation and numerical operations required during feature
engineering and preprocessing.

6. Matplotlib and Seaborn

These libraries were used for data visualization:

 Plotting model performance graphs (accuracy curves, confusion matrices).

 Visualizing class distributions and image samples.
 Assisting in exploratory data analysis (EDA).

7. Jupyter Notebook
Jupyter Notebook served as the primary environment for development and experimentation,
offering:

 An interactive platform for code development.

 Easy debugging and visualization.
 Rapid prototyping and iterative model building.

8. Google Colab
For heavy model training tasks, Google Colab provided free GPU access, enabling faster training of
CNN models without relying on local hardware.

CHAPTER - 4

SYSTEM DESIGN

Dept of CSE-AIML, SoET, CMRU 2024-25 9

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

The design of the AI Assisted Skin Disease Detection System focuses on modularity, scalability, and
ease of user interaction. The system combines machine learning-based classification with a user-
friendly interface to provide preliminary diagnosis support.

1. System Architecture

The system follows a modular architecture consisting of the following main components:

 Data Collection and Preprocessing Module

 Feature Extraction and Transformation Module

 Machine Learning Model (Classification Layer)

 Prediction and Output Module

 User Interface Layer (Web App or Mobile App)

2. Workflow of the System

1. Input Stage

 Users provide inputs in two ways:

o Uploading a skin image (photo of the affected area).
o Filling out a symptom-based questionnaire.

2. Preprocessing Stage
 Image resizing, normalization, and noise removal.
 Text cleaning and keyword extraction for symptoms.

3. Feature Extraction Stage

 For images: Features are extracted using CNN models.

 For text: TF-IDF vectorization of symptom descriptions.

4. Prediction Stage
 The preprocessed input is passed through a trained classification model.

 The model predicts the most probable skin disease.

5. Recommendation Stage

 Based on the disease, recommended treatments or next steps (e.g., "consult dermatologist",
"use OTC medication") are suggested.

Dept of CSE-AIML, SoET, CMRU 2024-25 10

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

6. Output Stage

 Diagnosis results and recommendations are presented through the user interface.

3. Flow Diagram

4.

Modules Description

 Preprocessing Module:
o Handles noise removal, standardization of image sizes, and cleaning of symptom
descriptions.
 Classification Module:
o CNN model classifies images based on skin features.
o Text classification model maps symptoms to disease labels.
 Recommendation Engine:
Uses a simple content-based recommendation technique based on disease-treatment
mapping.
 User Interface:
Provides an interactive platform for patients to input symptoms or upload images and
view results.

5. Technologies Used

Dept of CSE-AIML, SoET, CMRU 2024-25 11

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

 Frontend: HTML/CSS/Flask/Django (for Web App) or Flutter (for Mobile App).

 Backend: Python, TensorFlow/Keras, Scikit-learn.
 Database: Firebase / SQL (for storing images, results, feedback).
 Hosting: Google Cloud / AWS (for scalability if deployed live).

CHAPTER - 5

SYSTEM ANALYSIS

The system analysis phase focuses on evaluating the technical components, performance, strengths,
limitations, and future scope of the AI Assisted Skin Disease Detection System.

5.1. Input Parameters

 Image Input:
o Skin lesion photographs captured under natural light.
o JPEG, PNG formats accepted.

 Text Input:

o Symptom descriptions entered by users (e.g., “red patches”, “itchy skin”, “dry flakes”).

Dept of CSE-AIML, SoET, CMRU 2024-25 12

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

5.2. Preprocessing Parameters

 Image Preprocessing:

o Resizing all images to 224×224 pixels.

o Normalizing pixel values between 0 and 1.
o Augmentation: Rotation, flipping, zooming (for model robustness).

 Text Preprocessing:

o Lowercasing.
o Removing special characters and stopwords.
o Tokenizing keywords for TF-IDF feature extraction.

5.3. Feature Extraction

 Image Features:

o Extracted using convolutional layers of pre-trained CNN models like ResNet-50.

 Text Features:

o Extracted using TF-IDF (Term Frequency-Inverse Document Frequency) vectorization.

o High-weighted keywords indicate important symptoms.

5.4. Classification and Prediction

 Model Type:

o CNN-based model for images (with fine-tuning using transfer learning).

o Logistic Regression / Random Forest for text inputs.

 Performance:

o Achieved accuracy >85% on test datasets.

o Precision and Recall balanced to reduce false positives.

 Validation:

o 10-fold cross-validation applied.

o Confusion matrix and ROC curve plotted to assess model performance.

Dept of CSE-AIML, SoET, CMRU 2024-25 13

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

5.5. Output Analysis

 Correct prediction of the top 1 or top 3 probable diseases.

 Suggestions for OTC (Over-The-Counter) medications or dermatologist consultation.
 Confidence score provided along with the prediction (e.g., “Psoriasis – 87% confidence”).

5.6. Strengths and Analysis

 High-speed predictions (<1 second for text input; <3 seconds for image input).
 Modular architecture – easily upgradable with new disease datasets.
 Suitable for both standalone offline usage and cloud deployment.
 Dual input system (text + image) increases usability.

5.7. Limitations

 Dependence on quality of input images; blurry or poor lighting reduces accuracy.

 Limited dataset size might affect model generalization for rare skin conditions.
 The system suggests probable diseases but does not replace clinical diagnosis.

5.8. Future Enhancements

 Expand database to cover a wider range of rare dermatological conditions.

 Implement multilingual support for symptom descriptions.

 Add Explainable AI (XAI) features to explain the model’s prediction reasoning.

 Integrate a chatbot for continuous user interaction.

 Enable integration with telehealth platforms for direct doctor consultations.

Dept of CSE-AIML, SoET, CMRU 2024-25 14

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

CHAPTER - 6

IMPLEMENTATION PLAN

This implementation plan outlines the key phases, tasks, tools, and timeline to successfully design,
build, and test the AI Assisted Skin Disease Detection System. The development process is divided
into multiple stages to ensure clarity, quality, and scalability.

🔹 Phase 1: Requirement Analysis and Dataset Preparation

Objectives:

 Define system goals and functional requirements.

 Identify target users and usage scenarios (patients, healthcare assistants).
 Collect suitable datasets (image datasets of skin conditions, symptom-text mappings).
 Clean and label the dataset appropriately.

Tools:

 Kaggle, DermNet, PH2 or HAM10000 datasets

 Excel / Pandas for data cleaning Duration: 1 week

🔹 Phase 2: Data Preprocessing and Feature Engineering

Objectives:

 Resize, normalize, and augment image datasets.

 Clean and tokenize textual symptom data.
 Extract features using CNN (for images) and TF-IDF (for text).
 Encode disease labels for classification.

Tools:

 OpenCV, NumPy for image preprocessing

 Scikit-learn (TF-IDF), LabelEncoder Duration: 1 week

Dept of CSE-AIML, SoET, CMRU 2024-25 15

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

🔹 Phase 3: Model Development and Training

Objectives:

 Build and train machine learning models:

o CNN or pre-trained ResNet for image classification.
o Logistic Regression or Random Forest for symptom classification.
 Validate model accuracy using cross-validation and test sets.
 Tune hyperparameters for best performance.

Tools:

 TensorFlow/Keras (for CNN)

 Scikit-learn (for classical ML)
 Google Colab for training Duration: 2 weeks

🔹 Phase 4: Prediction and Recommendation Engine

Objectives:

 Integrate prediction layer with input/output modules.

 Create a rule-based or content-based treatment recommendation engine.
 Map disease predictions to a list of suggested treatments.

Tools:

 Python dictionaries, JSON mappings, optional use of Surprise library Duration: 1 week

🔹 Phase 5: User Interface Development

Objectives:

 Design a user-friendly interface for input (image upload/text form) and output (diagnosis +
treatment).
 Use web technologies (Flask, Django) or mobile tools (Flutter).
 Connect frontend to prediction model.

Tools:

 HTML/CSS/Flask or Flutter Duration: 1 week

🔹 Phase 6: Testing and Evaluation

Objectives:

Dept of CSE-AIML, SoET, CMRU 2024-25 16

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

 Conduct unit testing of each module.

 Perform usability testing with mock users.
 Measure accuracy, response time, and usability score.

Tools:

 Pytest for backend, manual testing for frontend Duration: 1 week

🔹 Phase 7: Documentation and Final Report

Objectives:

 Prepare technical documentation.

 Compile user manual and final project report.
 Create presentation slides for demo.

CHAPTER - 7

CONCLUSION

The AI Assisted Skin Disease Detection System successfully demonstrates the potential of
Artificial Intelligence in enhancing healthcare accessibility and improving early diagnosis of skin
conditions. By integrating machine learning models for image and text-based inputs, the system
provides a quick, preliminary diagnosis that can guide patients toward appropriate treatment or
further medical consultation.

Dept of CSE-AIML, SoET, CMRU 2024-25 17

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

The project addressed key challenges like symptom interpretation, image classification, and real-time
prediction through the use of advanced tools such as Convolutional Neural Networks (CNNs), TF-
IDF, and Logistic Regression. The modular design ensures scalability, allowing easy future
expansion to include more diseases, languages, and deployment platforms.

Although the system cannot fully replace the expertise of a trained dermatologist, it serves as a
valuable supportive tool that can help bridge the healthcare gap, especially in remote and
underserved areas. With further enhancement and validation against larger datasets, the system can
be integrated into telemedicine platforms, providing reliable skin disease screening at scale.

In conclusion, the project highlights how AI-driven solutions can make healthcare smarter, faster,
and more inclusive, aligning with the broader vision of digital transformation in the medical field.

REFERENCES
1. Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., & Thrun, S. (2017).
Dermatologist-level classification of skin cancer with deep neural networks. Nature, 542(7639),
115–118. [Link]
2. Tschandl, P., Rosendahl, C., & Kittler, H. (2018). The HAM10000 dataset, a large collection of
multi-source dermatoscopic images of common pigmented skin lesions. Scientific Data, 5,
180161. [Link]

Dept of CSE-AIML, SoET, CMRU 2024-25 18

 “AI ASSISTED SKIN DISEASE DETECTION SYSTEM”
<< 4AIML2061 >> – << SEAI >>

3. Codella, N. C., Nguyen, Q. B., Pankanti, S., & Smith, J. R. (2017). Deep learning ensembles for
melanoma recognition in dermoscopy images. IBM Journal of Research and Development,
61(4/5), 5:1–5:15. [Link]
4. Han, S. S., Kim, M. S., Lim, W., Park, G. H., Park, I., & Chang, S. E. (2018). Classification of
the clinical images for benign and malignant cutaneous tumors using a deep learning algorithm.
Journal of Investigative Dermatology, 138(7), 1529–1538.
[Link]
5. Brinker, T. J., Hekler, A., Utikal, J. S., Grabe, N., Schadendorf, D., Klode, J., ... & von Kalle, C.
(2019). Skin cancer classification using convolutional neural networks: Systematic review.
Journal of Medical Internet Research, 21(10), e11936. [Link]
6. Kawahara, J., BenTaieb, A., & Hamarneh, G. (2016). Deep features to classify skin lesions. 2016
IEEE 13th International Symposium on Biomedical Imaging (ISBI), 1397–1400.
[Link]
7. Rajpurkar, P., Irvin, J., Ball, R. L., Zhu, K., Yang, B., Mehta, H., ... & Ng, A. Y. (2018). Deep
learning for skin disease diagnosis: A review. arXiv preprint arXiv:1809.02639.
[Link]
8. Shrivastava, R., Jaiswal, A., & Rathore, Y. K. (2021). Skin disease detection using deep learning
and image processing. Materials Today: Proceedings, 47, 2473–2479.
[Link]
9. Mahbod, A., Schaefer, G., Wang, C., & Ecker, R. (2020). Transfer learning using a multi-scale
and multi-network ensemble for skin lesion classification. Computer Methods and Programs in
Biomedicine, 193, 105475. [Link]
10. World Health Organization. (2022). Skin diseases. WHO Fact Sheets. Retrieved from
[Link]

Dept of CSE-AIML, SoET, CMRU 2024-25 19