Skin Cancer Prediction Using CNN

Welcome to the Skin Cancer Prediction project! This repository contains a Convolutional Neural Network (CNN) model designed to predict skin cancer using deep learning techniques.

Overview

This project aims to build an accurate skin cancer detection model utilizing CNN architecture. By leveraging data augmentation and tuning hyperparameters, the model achieves a classification accuracy of 82%. This can aid in early detection and improve patient outcomes in dermatology.

Features

• CNN Architecture: Designed with 2D convolutional layers and max pooling to effectively extract features from images.
• Data Augmentation: Implemented techniques to enhance generalization and robustness of the model.
• Performance: Achieved an accuracy of 82% on the validation dataset.

Project Structure

• Skin_Cancer_Prediction.ipynb: Jupyter Notebook containing the model implementation, training, and evaluation.
• Skin_Data/Cancer_Non_Cancer: Directory containing the dataset used for training and testing the model.
• requirements.txt: List of required Python packages.

Technologies Used

• Python
• TensorFlow
• Keras
• NumPy
• Pandas
• Matplotlib

Installation

To set up the project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/TejasWaghmare18/Skin-Cancer-Prediction.git
   

2. Navigate to the project directory:

   cd Skin-Cancer-Prediction
   

3. Install the required dependencies:

   pip install -r requirements.txt
   

Model Configuration

• Batch Size: 7
• Loss Function: Binary cross-entropy
• Hyperparameters: Varying stride, filter sizes, and max pooling sizes were used to optimize performance.

Comparative Study