PaperBanana

pipeline illustrations

• Meet publication standards for top venues like NeurIPS, ICML, ACL, and more
• Support complex system visualizations including multi-agent interaction

flows and neural network architectures

📈 Statistical Chart Creation

• Uses code-based rendering (Python Matplotlib) rather than pixel generation

to ensure data accuracy

• Eliminates "numerical hallucination" problems common in general AI image

generators

• Guarantees mathematical precision for every data point in charts
• Supports bar charts, line plots, scatter plots, and other academic chart

types

• Generates publication-ready visualizations directly from raw JSON/CSV data

🎓 Educational Infographic Design

• Create scientifically accurate and highly readable educational illustrations
• Perfect for supplementary materials, tutorials, and course handouts
• Simplify complex concepts and enhance academic communication

✨ Aesthetic Enhancement

• Elevate hand-drawn sketches to professional quality
• Apply academic aesthetic standards for color, typography, and layout
• Ensure visual presentations meet academic publication requirements

--- 🔬 How It Works - Five-Agent Closed-Loop Architecture

PaperBanana employs a unique multi-agent collaboration system:

1. Retriever: Locates reference visual samples from a curated library 2. Planner: Translates textual descriptions into structured visual layouts 3. Stylist: Synthesizes aesthetic guidelines for color and typography 4. Visualizer: Renders illustrations using the Nano-Banana-Pro model 5. Critic: Conducts three rounds of self-reflection to catch errors and iteratively optimize

This closed-loop verification mechanism ensures generated figures are both beautiful and accurate.

--- 💡 Why Choose PaperBanana?

✅ Built for Academia: Specifically designed for research publication needs, not a general image generator ✅ Data Precision: Code-driven chart generation avoids common AI numerical errors ✅ Time-Efficient: Reduce hours of illustration work to just minutes ✅ Open Source: Licensed under CC BY-SA 4.0 International ✅ Research-Backed: Built on cutting-edge research published on arXiv

--- 📦 Extension Features

After installing this extension, you can:

• One-Click Access: Click the toolbar icon to instantly jump to the

PaperBanana platform

• First-Time Onboarding: Automatically opens a feature introduction page on

installation

• Seamless Integration: No need to remember URLs—start creating academic

figures anytime, anywhere

--- 👥 Who Should Use This?

• 📝 Graduate students and PhD candidates writing papers
• 🎓 Professors and instructors creating course materials
• 🔬 Researchers preparing conference presentations
• 📊 Data scientists needing data visualization
• 📖 Technical writers authoring documentation

--- 🔗 Related Resources

• Official Website: https://paper-banana.org
• GitHub Repository: https://github.com/dwzhu-pku/PaperBanana
• Research Paper: https://arxiv.org/abs/2601.23265
• License: CC BY-SA 4.0 International

--- 🔒 Privacy & Security

This extension collects no user data and serves solely as a quick access tool to the PaperBanana website. No additional permissions required—completely transparent and secure.

--- 🌟 Get Started

1. Install the extension 2. Click the PaperBanana icon in your browser toolbar 3. On the platform, input your paper description, context, and figure caption 4. Let AI generate publication-ready academic illustrations for you

Make academic illustration creation simple, fast, and professional!

--- 🏆 Technical Highlights

• Multi-Agent Framework: Five specialized AI agents working in harmony
• Code-Based Rendering: Matplotlib backend ensures numerical accuracy
• Iterative Refinement: Three-round critic feedback loop
• Academic-First Design: Tailored for research publication standards
• Flexible Input: Supports methodology diagrams, statistical plots, system

architectures, flowcharts, and poster assets

--- Perfect for researchers who value both speed and quality in their academic publications.