title	LUNG AI
emoji	🫁
colorFrom	green
colorTo	gray
sdk	gradio
sdk_version	6.5.1
app_file	app.py
pinned	false

Project Overview

LungAI is a deep learning project aimed at detecting and classifying lung cancer from CT scan images. This model uses a ResNet50 architecture with transfer learning to differentiate between cancerous and non-cancerous lung tissue, as well as classify specific types of lung cancer.

Model Performance

This model was trained for 50 epochs on the dorsar/lung-cancer dataset, achieving the following performance on the test set:

• Final Test Accuracy: 83.73%

• Best Validation Accuracy: 81.94%

Repository Structure

• Architecture/: Contains the core model script.

  • architecture.py: Defines, trains, and evaluates the ResNet50 model.

• Processed_Data/: (Not in repository) Directory created by the preprocessor.

• app.py: Runs the Gradio web demo.

• preprocess.py: Downloads and organizes data from Hugging Face.

• requirements.txt: List of Python dependencies.

• lung_cancer_detection_model.pth: (Generated) PyTorch weights.

• lung_cancer_detection_model.onnx: (Generated) ONNX model for deployment.

Data

The dataset is sourced from Hugging Face and is downloaded by the preprocess.py script:https://huggingface.co/datasets/dorsar/lung-cancer

Setup and Usage

Step 1: Install DependenciesFirst, ensure you have Python installed. Then, install the required Python libraries using the following command:

Bash

pip install -r requirements.txt

Step 2: Log in to Hugging FaceTo avoid rate-limiting errors when downloading the dataset, log in using the Hugging Face CLI:

Bash huggingface-cli login

Step 3: Download and Preprocess DataRun the preprocessing script. This will download the dataset, map the classes, and save them in the correct Processed_Data directory.

Bash

python preprocess.py

Step 4: Train the ModelRun the training script. This will train the model using your GPU, print the progress, and save the final .pth and .onnx model files.

Bash

python Architecture/architecture.py

**Step 5: Run the Model (Demo)**Run the Gradio application to see your trained model in action.

Bash

python app.py

This will provide a local URL to open in your browser.

Notes

• The preprocess.py script automatically creates the Processed_Data directory with train, validation, and test subdirectories.

• The architecture.py script is set up to find these specific folders.

Contributing

If you would like to contribute to this project, please fork the repository and submit a pull request. We welcome improvements, bug fixes, and new features.