Kidney Disease Classification

End-to-end kidney disease classification by finetuning VGG-16. Try it on Streamlit cloud

Built With

• Tensorflow
• DVC
• MLFLOW
• FastAPI
• Streamlit
• Containerization with Docker

Dataset

Kidney CT scan Kaggle Dataset

Classes: cyst, tumor, stone, and normal

How to run

• STEP 01- Clone the repository

git clone https://github.com/dahshury/Kidney-Disease-Classification-DL-Project

• STEP 02- Create a conda environment after opening the repository (optional)

conda create -n kidney python=3.8 -y

conda activate kidney

• STEP 03- install the requirements

Installing dependancies from requirements.txt

pip install -r requirements.txt

Dependancies for training using GPU (optional):

conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0

To Run prediction using the pretrained model

run the following command in the terminal for the streamlit web app

python app2.py

OR alternatively, you can use the html app

python app.py

After launching either app, open the following link in the browser

localhost:8080

You can now upload an image of a desired category, and click predict for the result.

Results

trainig loss: 0.5373 - accuracy: 0.9007 - val_loss: 2.3800, val_accuracy: 0.7181

Contact Me

Linkedin

Acknowledgements

The project was guided by DSwithBappy Youtube channel.