This project aims to develop a machine-learning model based on image data to predict skin cancer. Inspired by the research paper titled "Skin Cancer Prediction Using Deep Learning Techniques", our project builds upon the proposed techniques to create a predictive system with practical applications.
The dataset used in this project comprises two main files: HAM10000_metadata.csv
, containing metadata about the skin lesion images, and hmnist_28_28_RGB.csv
, which provides RGB image data in CSV format. This dataset, available on Kaggle, consists of 10,000 images, making it a substantial resource for training and evaluating our predictive model.
The model
directory houses the code related to the machine learning model. Within this directory, you'll find model.ipynb
, a Jupyter Notebook containing the code for model training and evaluation, and skin.h5
, which stores a pre-trained model for skin cancer prediction.
The static
directory contains static files utilized by the web application. This includes assets such as images and icons, CSS files for styling the application interface, JavaScript files for client-side functionality, and any third-party libraries or frameworks required for the application.
The tests
directory includes files for testing the application. These tests ensure the functionality and reliability of the system, helping to identify and resolve any potential issues.
The web application provides an accessible interface for users to interact with the skin cancer prediction system. It consists of several HTML pages, including the initial landing page (first.html
), the main application page (index.html
), the login page (login.html
), and the prediction page (prediction.html
). The Flask application code for serving the web application is contained within app.py
.
To utilize the skin cancer prediction system, ensure all dependencies listed in requirements.txt
are installed. Then, run the Flask application using python app.py
, and access the application through a web browser using the provided URLs.
Through model training and evaluation, we achieved an accuracy of 97.56% in predicting skin cancer. This high level of accuracy demonstrates our predictive model's effectiveness in identifying potential skin cancer instances.
This project was a collaborative effort by a team of dedicated individuals, including:
- Ansh Prakash
- Tanisha Chandani
- AmulGupta
- Sanskriti
- Sanskar Sinha
- Abhishek Kumar
- Aviral Pandey
- Munish Thakur
- Harsh Pratap Singh
Make sure to have the following dependencies installed:
pip install scikit-learn==1.2.2
pip install flask==2.3.2
pip install matplotlib==3.7.1
pip install pandas==2.0.0
pip install numpy==1.23.5
pip install tensorflow==2.12.0
Make sure to replace pip install
with pip3 install
if you're using Python 3. x.
The inspiration for our project and the base paper we referred to as "Skin Cancer Prediction Using Deep Learning Techniques". You can find the paper here.