This project implements a CNN trained by CIFAR/10 in order to implement image recognition to create a telephone like game. The players draw the images and the machine attemps to identify the image, passing it on to the next user.

This project uses the PyTorch Chainguard image as a base image. This image is free of vulnerabilities (zero CVEs), guaranteeing that external actuators cannot exploit vulnerabilities. However, two dependencies were added for the development of the GUI: streamlit and streamlit-drawable-canvas

# Clone the repository
git clone https://github.com/Rodrigo-Munoz-Salas/AI_Image_Recognition.git

# Build the Docker image (from Chainguard images)
docker build . -t "ai-image-recognition"

# Run the docker container
docker run -d -p 8501:8501 ai-image-recognition

Go to your local host at 8501 port: localhost:8501

Select the number of players, click on Start, and begin drawing! Once you finish your drawing, click on Submit Drawing and wait for the program to give you its prediction!

# Stop the Docker container
docker stop CONTAINER_ID