Paper2Poster

One-Minute Overview#

Paper2Poster is a multi-agent system that automatically converts academic papers from PDF files into editable PowerPoint posters. It uses a top-down, visual-in-loop approach with multiple AI agents working together to extract key paper content and create visually appealing posters. Researchers, students, and academics can use this tool to quickly transform complex academic papers into clear, concise poster formats suitable for conferences, presentations, and sharing.

Core Value: Significantly reduces academic poster creation time by converting lengthy papers into visually clear posters while maintaining academic accuracy.

Quick Start#

Installation Difficulty: Medium - Requires Python environment, multiple dependencies, and API key configuration

# Install dependencies
pip install -r requirements.txt
# Install LibreOffice
sudo apt install libreoffice
# Install poppler
conda install -c conda-forge poppler

Is this suitable for my scenario?

• ✅ Rapid academic conference poster generation: Convert papers to posters in one click, saving hours of manual design time
• ✅ Multiple model options: Choose between GPT-4o or open-source models based on performance needs and budget
• ❌ Precise visual customization: Currently focuses more on automatic content extraction than design flexibility
• ❌ Completely offline usage: Some features require API access

Core Capabilities#

1. Multi-Agent System#

• PosterAgent uses a top-down, visual-in-loop multi-agent architecture that converts PDF papers into editable PPTX posters Actual Value: Intelligently coordinates multiple AI agents for分工合作, ensuring comprehensive and well-structured poster content

2. Flexible Model Combinations#

• Supports various LLM/VLM combinations including GPT-4o and local open-source models (Qwen-2.5-7B-Instruct) Actual Value: Choose different model combinations based on performance requirements and budget, balancing effectiveness and cost

3. Intelligent Content Extraction#

• Automatically extracts core content from papers including abstract, methods, results, and other key sections Actual Value: Eliminates the tedious process of manually reading lengthy papers, quickly obtaining and presenting the most valuable information

4. Automated Logo Integration#

• Automatically searches and adds institutional and conference logos, supporting both local and online search Actual Value: Enhances poster professionalism and brand recognition, saving time on manual logo searching and adding

5. YAML Style Customization#

• Customizable poster appearance through YAML configuration files for both global and per-poster settings Actual Value: Provides flexibility for users to adjust poster appearance according to conference requirements or personal preferences

Tech Stack & Integration#

Development Language: Python Main Dependencies: vLLM (for open-source model deployment), Libreoffice (for document processing), poppler (for PDF processing), OpenAI API (for GPT-4o), Google Search API (for logo search) Integration Method: Command-line tool, API, Docker container

Maintenance Status#

• Development Activity: Very active, with multiple updates in 2025 adding new features
• Recent Updates: Recently added Gradio demo, Docker support, automatic logo search, and other features
• Community Response: Project has been accepted to NeurIPS 2025 Dataset and Benchmark Track, indicating academic recognition

Documentation & Learning Resources#

• Documentation Quality: Comprehensive
• Official Documentation: Detailed installation and usage guides available in the README
• Sample Code: Command-line examples provided for various model combinations

Commercial & Licensing#

License: Unknown (not explicitly stated in the project)

• ✅ Commercial: Unknown
• ✅ Modification: Unknown
• ⚠️ Restrictions: Unknown