PaperBanana: Automate Research Figures with AI – Open Source Tool
Introduction
Academic writing is often bogged down by the tedious process of creating and formatting figures that match the rigor and aesthetics of a research paper. PaperBanana tackles this problem head‑on by leveraging modern machine learning techniques to automatically generate high‑quality illustrations from raw data. Built as an open‑source GitHub repository, it offers a community‑friendly way to streamline figure creation for AI scientists and researchers across disciplines.
What Is PaperBanana?
PaperBanana is an end‑to‑end web application written primarily in JavaScript, HTML, and CSS. Its core idea is simple: turn data into polished visual assets with minimal human intervention. The project comprises:
- Data ingestion pipeline – Users upload CSV, JSON, or even markdown tables.
- Template engine – A library of pre‑designed figure templates (scatter plots, bar charts, heatmaps, neural‑network diagrams, etc.).
- AI backend – A lightweight inference model (trained on a curated dataset of scientific figure styles) selects the best template and auto‑adjusts layout, color, and annotation.
- Export options – Images can be saved as PNG, SVG, or LaTeX-friendly figures for seamless integration into manuscript LaTeX files.
The repository is still in active development: a new stable release is expected in ~2 weeks, and a detailed tutorial is on its roadmap.
Core Features
| Feature | Description |
|---|---|
| Auto‑Template Matching | The system scans your dataset, infers the type of plot that best represents it, and applies a matching template. |
| Intelligent Labeling | Axis titles, legends, and statistical annotations are generated automatically based on data characteristics. |
| Style Consistency | Choose from a palette that mimics common journal styles (IEEE, Nature, CVPR) to maintain visual coherence across a manuscript. |
| Customizable Parameters | Fine‑tune plot size, color schemes, and LaTeX captions via simple configuration files or a UI overlay. |
| CLI & API | Advanced users can integrate PaperBanana directly into CI pipelines or Jupyter notebooks by invoking its lightweight command‑line interface. |
How It Works
- Upload Your Data – Use the web UI or the CLI to submit raw data.
- Data Profiling – PaperBanana performs a quick statistical analysis (mean, variance, distribution shape) to determine the most appropriate visualization.
- Template Selection & Rendering – The AI engine selects a template and renders the figure using D3.js and Canvas.
- Optimization – The generated image is automatically trimmed, annotated, and scaled for the requested output format.
- Download – Grab your figure in your preferred format and embed it in your manuscript.
All of this happens in under a minute for most typical datasets, dramatically cutting the time researchers spend on figure preparation.
Getting Started
- Clone the Repo
git clone https://github.com/dwzhu-pku/PaperBanana.git cd PaperBanana - Install Dependencies
npm install - Serve the App
npm start - Open in Browser – Navigate to
http://localhost:3000and try uploading a sample CSV.
Alternatively, the project contains a Dockerfile for containerized deployments, making it easy to host PaperBanana on any cloud platform.
Community & Contributing
PaperBanana invites researchers, data scientists, and developers to contribute. The repo is under a Creative Commons Attribution‑ShareAlike 4.0 license, encouraging community reuse. If you wish to:
- Add new figure templates (e.g., violin plots, Sankey diagrams)
- Improve the AI model’s accuracy with more training data
- Write blog posts or tutorials about using the tool
Simply fork the repo, implement your changes, and open a pull request. The project team will review and merge contributions following the repository’s contribution guidelines.
Future Roadmap
- Release 1.0 – Full documentation, stable CLI, and pre‑trained models.
- Community Templates – A marketplace for user‑submitted figure styles.
- Multi‑language Support – Integrations for Python, R, and Julia.
- Batch Processing – One‑click conversion of large datasets into a series of figures.
Keep an eye on the GitHub issues page for the latest feature requests and bug reports.
Conclusion
PaperBanana democratizes the creation of publication‑ready figures. By automating routine tasks, it frees researchers to focus on analysis and interpretation rather than formatting. With its open‑source foundation, community‑driven templates, and AI‑powered rendering, PaperBanana is poised to become a staple in the scientific workflow toolbox. Give it a try and see how it can accelerate your research publishing process today!