laansdole / image-caption-pipeline Goto Github PK

View Code? Open in Web Editor NEW

FastAPI Production Setup with pipeline for HuggingFace image caption model

Python 72.67% Dockerfile 27.33%

image-caption-pipeline's Introduction

FastAPI pipeline for Image Captioning Model

This project uses Python 3.10. Remember to run below commands in the root directory of your project where the pyproject.toml file is located. Here are the main elements you need to know to get started:

Model: nlpconnect/vit-gpt2-image-captioning

Virtual Environment

Note that for this the package venv is to be used.

In short, a virtual environment will help us manage an isolated version of python interpreter. And so too installed packages. In this way. Different project will not have to depends on the same packages installation and have to conflict. Read the link above explain and show it well.

Build on local

Docker Build

docker compose up --build

Poetry or Pip

# Create a virtual environment, navigate to your root directory and run:
python -m venv .venv
# Activate the environment, use:
source .venv/bin/activate
pip install -r requirements

In a Python project managed with Poetry, dependencies are listed in a pyproject.toml file, make sure that [tool.poetry] name = "bloomsage-backend":

poetry install --no-root

To add additional dependencies/packages to your project, use:

With poetry

poetry add [dependencies/packages]

With pip

pip install [dependencies/packages]

Once you have done with your session

# To export the requirements.txt
pip install -r requirements
poetry export -f requirements.txt --output requirements.txt --without-hashes --without=dev
# Deactivate the environment once done, use:
deactivate