OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Litepali

LitePali is an open-source Python library that indexes and searches document images with a ColPali-inspired vision-language model architecture. It supports direct image input, batch processing, text-query retrieval, and saving or loading an index.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Litepali — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use LitePali when your retrieval corpus consists of document images and you want a Python-based semantic search component. For PDF collections, pair it with a separate page-image conversion step.

Who it is for

  • Developers adding image-based document retrieval to Python applications.
  • Researchers working with scanned or image-native document collections.
  • Teams building multimodal retrieval-augmented generation pipelines around document images.

Documented uses

  • Index document-page images with document IDs, page IDs and optional metadata.
  • Search an indexed image collection using a text query and retrieve ranked image results.
  • Save an index for later loading in another LitePali instance.
  • Add multiple image files through a batch-oriented workflow.
Before you choose

Requirements & limits

The project is published as a public GitHub repository under the MIT license. The README provides pip installation instructions.

  • The library accepts images directly and does not provide the documented PDF conversion stage.
  • Its retrieval approach is based on ColPali and combines visual and textual document understanding.
  • The README specifies Python 3.7+ and PyTorch 1.8+.
  • Index saving and loading are included in the documented usage example.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Index and search a collection of document images

Add image files to LitePali, process them, and query the resulting collection.

Before you start

  • Python 3.7+ and PyTorch 1.8+ are listed in the README.
  • Install the library with pip.
  • Have document images available as local file paths.

Steps

  1. Install LitePali with the command below.

    Documented command
    pip install litepali

  2. Import LitePali and ImageFile, then initialize a LitePali instance.

  3. Add each image as an ImageFile, supplying its path, document ID, page ID, and optional metadata.

  4. Call process(), then search with a text query using search(query, k=5) or another chosen result count.

Expected outcome

Search results include the image and its score, as shown in the README example.

Guide limitations

  • LitePali works directly with images; PDF page conversion must be handled separately.
  • The guide reflects the README example and is not a report of an independently run test.
Documentation 1 ↗
Source-based practical guide

Save and reload a LitePali index

Preserve a processed image index and load it into another LitePali instance.

Before you start

  • A LitePali instance with an index to save.
  • A destination path for the saved index.

Steps

  1. Call save_index() with the path where the index should be saved.

  2. Initialize another LitePali instance when loading the index.

  3. Call load_index() with the saved index path.

Expected outcome

A new instance loads the saved index for subsequent use.

Guide limitations

  • The README shows the save and load calls but does not specify index format or storage-backend options.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
s-emanuilov/litepali · Official homepage title · retained in its source language
How we review sourcesSuggest a correction