/imagercg-waiter

Build and Deploy Custom Docker Images for Object Recognition, Towards AI, 2023.

Primary LanguagePythonMIT LicenseMIT

Image Recognition Waiter

This application serves a deep learning image classification model that recognizes what object is present in an image. It accepts images from the user, makes request to an API endpoint that makes a prediction, and shows results in a frontend UI.

It demonstrates use-cases of different tools such as PyTorch, FastAPI, Gradio and Docker.

A detailed writeup is published in Towards AI!


Model Output: `king penguin: 0.99`.

Usage

To launch the application, run:

git clone https://github.com/hasibzunair/imagercg-waiter
cd imagercg-waiter/backend
sh deploy.sh

The app is live in http://0.0.0.0:7860. Upload images to make a prediction, or simply use the examples! For details on how the frontend and backend components were built, see respective folders.

Note

I did this project after completing Docker for the Absolute Beginner - Hands On - DevOps.

Todos

  • Google cloud run for backend
  • Docker compose
  • Kubernetes (make some pods lol!)

References

Also see learn-docker.