Vision Text Extractor

Last updated on Oct 19, 2025
Code Documentation
Vision Text Extractor - AI-powered OCR with multiple provider support

Features

The Vision Text Extractor is a powerful CLI tool that combines multiple AI providers for optimal text extraction from images and documents. Whether you need privacy-first local processing or high-accuracy cloud-based solutions, this tool has you covered.

Key Capabilities:

  • Multiple AI Providers: Choose from local SmolVLM/LLaVA models or cloud-based OpenAI GPT-4o
  • Privacy-First Design: Local processing keeps your sensitive data on your machine
  • Flexible Input Options: Process local files or images from web URLs
  • Custom Prompts: Extract specific information with tailored extraction prompts
  • Easy Setup: One-command installation using Pixi package manager

Quick Start

Getting started with Vision Text Extractor is simple:

# Clone and install
git clone https://github.com/udit-asopa/vision-text-extractor.git
cd vision-text-extractor
pixi install

# Setup
pixi run setup-smolvlm    # Hugging Face SmolVLM (~2GB)
pixi run setup-ollama     # Ollama LLaVA (~4GB)

# Quick demo
pixi run demo-ocr-huggingface

# Use with your images refer this for quick Commands: 
# https://github.com/udit-asopa/vision-text-extractor/tree/main?tab=readme-ov-file#-quick-commands
pixi run ocr_llm path/to/your/image.jpg
OCR Computer Vision Object Detection Text Extraction Document Processing AI Artificial Intelligence Machine Learning Deep Learning Vision-Language Models Multimodal AI Large Language Models (LLMs) Natural Language Processing OpenAI LLaVA SmolVLM Foundation Models Python Python Development CLI Tools Command Line Interface Open Source GitHub Software Development API Development Model Deployment Edge AI Local Processing Cloud AI Services Privacy Protection Data Privacy Document Digitization Automation Business Documents Enterprise Applications Technical Writing Innovation Problem Solving
Udit Asopa
Udit Asopa
Remote Sensing Data Specialist (official title: SAR Remote Sensing Engineer)

I specialize in SAR and GIS-based Earth observation workflows for environmental monitoring, disaster response, and scientific analysis. With a focus on automation, reproducibility, and applied geospatial intelligence, I contribute to building scalable, data-driven solutions.

Previous

Related