Card Prediction Application

A simple machine learning application that uses a Convolutional Neural Network (CNN) models trained on images of playing cards and predicts the card type from an uploaded image.

Models are trained in notebook files, and the trained model is saved in the model directory. The application uses TensorFlow for image processing, prediction and Flask for the web interface.

Setup Instructions

Prerequisites

• Conda (Anaconda or Miniconda)
• Python 3.9
• NVIDIA GPU (optional, for faster training and inference)
• CUDA toolkit and cuDNN (if using GPU)

Cloning the Repository

Clone the repository and navigate to the project directory:

  git clone https://github.com/pirrozani/Card-Prediction-Application.git && cd Card-Prediction-Application

Downloading the Dataset

Note

This step is necessary only if you want to train the model yourself. If you want to use the pre-trained model, you can skip this step and directly run the application.

To download the dataset, follow these steps:

1. Go to the Kaggle dataset page Cards Image Dataset-Classification.
2. Click on the "Download" button to download the dataset as a ZIP file.
3. Extract the ZIP file to a directory of your choice.
4. Place the extracted dataset in the data directory of this project.

Creating a Python Environment for TensorFlow GPU using Conda

Follow these steps to set up a Python environment with GPU support for TensorFlow:

# Create a new conda environment with Python 3.9
conda create -n tf_gpu python=3.9

# Activate the environment
conda activate tf_gpu

# Install CUDA toolkit and cuDNN for GPU acceleration
conda install -c conda-forge cudatoolkit=11.2 cudnn=8.1.0

# Upgrade pip
python -m pip install --upgrade pip

# Install required packages from `requirements.txt`
pip install -r requirements.txt

Running the Application

After setting up the environment, you can run the application by:

# Make sure you have activated the tf_gpu environment
conda activate tf_gpu

# Run the Flask web server
python main.py

Using Docker

You can run the application in a Docker container instead of using conda. This is particularly useful if you want to avoid dependency issues or if you prefer a containerized environment.

Prerequisites for Docker

• Ensure Docker and Docker Compose are installed on your machine.

Running with Docker

Important

Make sure you are in the project directory where the docker-compose.yml file is located.

Build the Docker image and run the services using Docker Compose command.

GPU Support

For GPU acceleration, you can use the following command to build and run the application:

  docker-compose -f docker-compose.yml -f docker-compose.gpu.yml up --build

CPU Only

If you do not have a GPU or prefer to run the application without GPU support, you can use the following command:

  docker-compose -f docker-compose.yml up --build

In order to check if GPU acceleration is available, you can run:

  docker-compose exec card-prediction nvidia-smi
  # It will display the GPU information if available

To check if the services are running, you can use:

  docker ps

Note

• The Docker setup automatically mounts the models and uploads directories for persistence
• GPU acceleration is enabled by default if NVIDIA Container Toolkit is available
• The container uses CUDA 11.8 with cuDNN 8 for optimal TensorFlow GPU performance

Stop the Application

You can stop the services using:

  docker-compose down

Access the Application

Once the server is running, you can access it through your web browser no matter if you used conda or Docker installation. Open your web browser and navigate to:

http://localhost:5000