Edit-Banana

Edit Banana: A framework for converting statistical formats into editable.

What is Edit-Banana?

Edit Banana: A framework for converting statistical formats into editable.

Overview of Edit-Banana

What it is

Edit Banana is a framework for converting statistical formats into editable assets. It transforms static images (PNG, JPG, BMP, TIFF, WebP) into fully manipulatable DrawIO (XML) diagrams. The project is developed by BIT-DataLab and is available as open-source software under the AGPL-3.0 license. The web-based service is available at editbanana.net or editfigure.com, and the GitHub repository may lag behind the web platform in features and performance.

Key features

  • Advanced Segmentation: Uses a fine-tuned SAM 3 (Segment Anything Model 3) for diagram element segmentation.
  • Fixed Multi-Round VLM Scanning: An extraction process guided by Multimodal LLMs.
  • Text Recognition: Includes local OCR (Tesseract) for text localization, Pix2Text for mathematical formula recognition and LaTeX conversion, and a crop-guided strategy for high-resolution text/formula region extraction.
  • User System: Registration provides 10 free credits; pay-per-use credit system prevents resource abuse; multi-user concurrency with a Global Lock mechanism for thread-safe GPU access and an LRU Cache for image embedding persistence.
  • DrawIO XML Generation: Merges spatial data from SAM3 and text OCR results to produce editable diagrams.

Use cases

  • Converting static diagrams (flowcharts, architecture diagrams, technical schematics, scientific formulas) into editable DrawIO files.
  • Human-in-the-loop modification of reconstructed diagrams, allowing manual repair and local saving.
  • Academic or professional scenarios where fixed-format figures need to be re-edited while preserving layout, color, element hierarchy, and text.

Getting started

See the repository README for detailed installation and setup instructions. The README provides steps for environment setup, cloning the repository, installing dependencies (including PyTorch with CUDA, SAM3, Tesseract OCR, and optional PaddleOCR, Pix2Text, and RMBG), and configuring the project.

License and pricing

The project is licensed under AGPL-3.0. The web service offers a free tier (10 free credits upon registration) and a pay-per-use credit system. Pricing details beyond this are not stated in the provided materials.

Based on the project's public README and repository metadata.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/bit-datalabedit-banana.svg)](https://yunodb.com/tools/bit-datalabedit-banana?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/bit-datalabedit-banana?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/bit-datalabedit-banana.svg" alt="Featured on YunoDB" height="20"></a>