Skip to content
This repository was archived by the owner on Jul 14, 2026. It is now read-only.

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

License: MIT Python NLP OCR


Overview

DocuSense is an intelligent document analysis platform that combines OCR, semantic search, summarization, and interactive Q&A across heterogeneous document types. It transforms static document collections into queryable knowledge bases — enabling natural language interaction with any document corpus.

The gap between storing documents and understanding them is what DocuSense closes.


Architecture

Document Input (PDF, Image, Text)
         │
         ▼
┌─────────────────┐
│  OCR Pipeline   │   Tesseract + preprocessing for scanned docs
│                 │   Text extraction for native PDFs
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Semantic Index │   Sentence Transformers embeddings
│                 │   Vector storage for similarity search
└────────┬────────┘
         │
    ┌────┴────┐
    ▼         ▼
┌────────┐ ┌──────────────┐
│Summarize│ │ Semantic Q&A │
│Pipeline │ │   Pipeline   │
└────────┘ └──────────────┘

Key Features

Feature Description
OCR Recognition Extracts text from scanned documents and images
Semantic Search Meaning-based search across entire document corpus
Auto Summarization Abstractive and extractive document summarization
Interactive Q&A Natural language queries answered from document context
Multi-Format PDF, images, plain text, and mixed document types

Tech Stack

Layer Technology
Embeddings Sentence Transformers
OCR Tesseract / pytesseract
NLP Transformers, spaCy
Backend Python
Language Python 3.10+

Getting Started

git clone https://github.com/royxlead/docusense-python.git
cd docusense-python
pip install -r requirements.txt
python app.py

Research Context

DocuSense explores the practical limits of dense retrieval and semantic search on heterogeneous real-world documents — including scanned PDFs where OCR quality directly affects retrieval performance. The semantic search layer uses bi-encoder architecture for efficient similarity computation across large document collections.


Related Work

  • CURA — Domain-specific RAG for medical documents
  • Auto-Researcher — Multi-agent academic paper analysis

About

AI-powered platform for intelligent document analysis, summarization, semantic search, and interactive Q&A using cutting-edge NLP technologies.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages