Kidney CT Scan Image Classification using VGG16

This project is an end-to-end deep learning application that uses the VGG16 model for transfer learning to classify kidney CT scan images. The dataset used is sourced from Kaggle. The project also incorporates MLflow and DagsHub for experiment tracking, and DVC for data versioning and pipeline management.

Experiment-Tracking

• Dagshub URL : https://dagshub.com/Sidd-77/kidney-disease-classification
• Mlflow URL: https://dagshub.com/Sidd-77/kidney-disease-classification.mlflow

Tech Stack

• Deep Learning: TensorFlow, Keras

• Transfer Learning: VGG16 model

• Experiment Tracking: MLflow, DagsHub

• Data Versioning: DVC, GIT

• Containerization: Docker

• Dataset: Kaggle CT Kidney Dataset

Data Pipeline

Model Evalutaion

Setup and Installation

1. Clone the repository:

git clone https://github.com/Sidd-77/kidney-disease-classification.git

2. Navigate to the project directory:

cd kidney-disease-classification

3. Install the required dependencies:

pip install -r requirements.txt

4. Run DVC pipeline:

dvc repro

Running the Application

1. To train the model, run:

python main.py

2. To start the Streamlit application, run:

streamlit run application.py

Docker

A Dockerfile is provided if you wish to build a Docker image.

1. Build the Docker image:

docker build -t <image-name>  .

2. Run the Docker container:

docker run -p 8501:8501 <image-name>

The application will be accessible at http://localhost:8501.