pyproject.toml

[tool.poetry]
name = "document-extraction-colpali-blip-2"
version = "0.1.0"
description = "This project is a multimodal document retrieval and question-answering system powered by ColPali and BLIP-2 models. It provides two modes of operation: a Command Line Interface (CLI) and a FastAPI-based REST server. In the CLI mode, users can query a folder or file of documents (images) and receive a response to their question directly in the terminal. In server mode, the FastAPI application exposes two endpoints: a health check and a document-query endpoint. The document-query endpoint allows users to upload documents and ask a question via a simple HTTP request, returning the generated response in real time. The system leverages ColPali for document retrieval and BLIP-2 for generating context-aware answers, making it suitable for retrieval-augmented generation (RAG) tasks on multimodal datasets."
authors = ["Jens Weber <jens@aip.engineer>"]
license = "MIT"
readme = "README.md"

[tool.poetry.dependencies]
python = "^3.12"
colpali-engine = ">=0.3.0,<0.4.0"
transformers = "^4.44.2"
torch = "^2.4.1"
pillow = "^10.4.0"
fastapi = "^0.114.1"
uvicorn = "^0.30.6"
python-multipart = "^0.0.9"
pdf2image = "^1.17.0"
torchvision = "^0.19.1"
torchaudio = "^2.4.1"
python-dotenv = "^1.0.1"


[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"