Pretrained model deployment with FastAPI and Docker.

Project description

Here I use pretrained EfficientNetB0 model from tensorflow.keras.applications for image classification. Example of images can be found in imgs folder.

How To Run

app.py is an entry point of the project

Run locally

1. Set up a virtual environment:

python -m venv .venv
.venv\Scripts\activate 
pip install -r requirements.txt

2. Run uvicorn app:app --port 8080
3. Open index.html in browser
4. Select any image and classify it

Example:

Run in a Docker container

The only thing you need to run this project in Docker is Docker Desktop installed on your computer.

1. Run .\deploy.ps1 in PowerShell
2. Open index.html in browser
3. Select any image and classify it

Additionally

• To create requirements.txt file use pip list --format=freeze > requirements.txt command to avoid weirdly looking paths