How to Build and Deploy an Image Recognition App using FastAPI and PyTorch ?

This is an image recognition application based on the FastAPI framework and PyTorch which uses pretrained DenseNet 121 model to detect the Image.

Tutorial

Installation

Run my Project

  git clone https://github.com/bhimrazy/Image-Recognition-App-using-FastAPI-and-PyTorch
  cd Image-Recognition-App-using-FastAPI-and-PyTorch
  python -m venv venv
  source venv/bin/activate
  pip install -r requirements.txt
  uvicorn main:app --reload

Image Recognition App using FastAPI and PyTorch: TODO

• [x] Create a virtual environment

• [x] Create FastAPI App

  • Install fast API
  • Install Uvicorn
  • Install Pytest
  • Install Jinja2
  • Install python-multipart
  • Install requests
  • Create a main file with some routes
  • Create a main test file to test the home page route

• [x] Pytorch Setup

  • Install torch & torchvision (use cpu version for small size)

• [x] Prediction

  • Create a predict post route
  • Create a file utils.py
  • Test predict route
  • Create some helper function
  • Put some test images inside static folder
  • Create a test to upload an image in predict route
  • Predict

• [x] Create a home page for prediction

  • Create an index.html file inside the templates directory
  • Setup template and static directory in the main app
  • Initial HTML for home page
  • Use Tailwind CSS cdn link for css
  • Google Fonts
  • Create a form to predict
  • update homepage route for prediction
  • Update UI of the page
  • Add Some javascript to autoload the image
  • Add logo and favicon
  • Add meta tags
  • Add response image for preview as base64 data

📚 RESOURCES:

◆ PyTorch: https://pytorch.org/tutorials/intermediate/flask_rest_api_tutorial.html ◆ FastAPI: https://fastapi.tiangolo.com

Author

• @bhimrazy