Edit-Banana
Edit-Banana is a Python framework that converts static diagrams (PNG, JPG, etc.) into editable DrawIO XML format. It uses SAM 3 for element segmentation, multimodal LLMs for content extraction, and OCR/Pix2Text for text and formula recognition, enabling users to modify previously non-editable diagrams.
Key facts
Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.
| Field | Value |
|---|---|
| Repository | BIT-DataLab/Edit-Banana |
| Owner | BIT-DataLab |
| Primary language | Python |
| License | AGPL-3.0 — OSI-approved |
| Stars | 5.4k |
| Forks | 361 |
| Open issues | 35 |
| Latest release | Unknown |
| Last updated | 2026-07-05 |
| Source | https://github.com/BIT-DataLab/Edit-Banana |
What Edit-Banana is
The system combines fine-tuned SAM 3 segmentation, parallel text extraction (Tesseract OCR + Pix2Text for LaTeX), and DrawIO XML generation. It includes a credit-based user system, LRU caching for embeddings, and concurrent processing via global GPU locks. A web demo runs at editbanana.net; GitHub code lags behind the live service.
Get the Edit-Banana source
Clone the repository and explore it locally.
git clone https://github.com/BIT-DataLab/Edit-Banana.gitcd Edit-Banana# follow the project's README for install & configurationNeed it deployed, integrated, or customized instead? DEV.co ships production installs.
Best use cases
Implementation considerations
- Requires CUDA-capable GPU and sufficient VRAM for SAM 3 + LLM inference; local deployment cost may be high.
- AGPL-3.0 license necessitates legal review before integration into commercial or proprietary systems; consider GPL exception or commercial license inquiry.
- Model weights (SAM 3) must be manually downloaded and placed in /models/; installation script provided but not verified for all environments.
- GitHub version lags live service; production use should clarify feature parity and consider adopting the web API (editbanana.net) instead.
- Credit-based system and user registration in place, but no pricing docs, quota limits, or SLA terms visible in README.
When to avoid it — and what to weigh
- Requires Latest Features — GitHub repo is stated to trail the live web service significantly; self-hosted deployments may lack recent improvements.
- Strict AGPL Compliance Burden — AGPL-3.0 requires distributing source code of derivative works; closed-source commercial products or SaaS without source disclosure face legal risk.
- GPU Resource Constraints — SAM 3 and multimodal LLM inference require CUDA; no clear CPU fallback or quantization guidance in README.
- Zero Production Stability Signal — No formal releases, minimal issue resolution rate (35 open issues, ~6 months active), and no documented SLAs or uptime guarantees.
License & commercial use
Licensed under AGPL-3.0. This is a strong copyleft license: any derivative work or service must disclose source code and provide it to users. Commercial use is possible but requires either open-sourcing derivatives or obtaining a separate commercial license from the authors (contact [email protected]).
AGPL-3.0 is incompatible with closed-source commercial products without explicit exception. SaaS deployment of Edit-Banana itself requires source code availability to end-users. Recommend obtaining written clarification or a commercial license from BIT-DataLab before building proprietary systems. README lists contact for 'commercial licensing' inquiry.
DEV.co evaluation signals
Editorial assessment — not user reviews. Directional, with an explicit confidence level.
| Signal | Assessment |
|---|---|
| Maintenance | Active |
| Documentation | Adequate |
| License clarity | Clear |
| Deployment complexity | High |
| DEV.co fit | Good |
| Assessment confidence | Medium |
No explicit security audit or hardening details provided. Multi-user system includes global GPU locks and LRU caching, but input validation, rate limiting, and prompt injection risks (LLM-based extraction) are not documented. File upload handling and data retention policies unknown. AGPL source availability aids security review but doesn't guarantee it has been done.
Alternatives to consider
Diagram.net (DrawIO) Native Reverse-Engineering
DrawIO offers manual import and export; no automation, but zero license/security friction and full control. Suitable for low-volume, high-accuracy needs.
Tesseract-based OCR + Custom XML Gen
Build a lightweight pipeline using open-source OCR (Tesseract) + custom logic; avoids LLM dependency, AGPL compliance, and GPU cost. Trade-off: no semantic understanding.
CloudConvert / Online Converters (Closed-Source SaaS)
Outsource diagram conversion; no deployment burden or AGPL concerns. Trade-off: privacy, cost per conversion, and vendor lock-in.
Build on Edit-Banana with DEV.co software developers
Test the live demo at editbanana.net. For integration, clarify AGPL licensing, GPU requirements, and feature parity with GitHub. Contact BIT-DataLab for commercial licensing.
Talk to DEV.coRelated open-source tools
Surfaced by semantic similarity across the DEV.co open-source index.
Related on DEV.co
Explore the category and the services that help you build with it.
Edit-Banana FAQ
Can I use Edit-Banana in a closed-source product?
Does the GitHub code match the online demo at editbanana.net?
What hardware do I need to run this locally?
How accurate is the formula recognition?
Software development & web development with DEV.co
From first prototype to production, DEV.co delivers software development services around tools like Edit-Banana. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across ai frameworks and beyond.
Evaluate Edit-Banana for Your Diagram Workflow
Test the live demo at editbanana.net. For integration, clarify AGPL licensing, GPU requirements, and feature parity with GitHub. Contact BIT-DataLab for commercial licensing.