Skip to content
Back to search

Paper2Poster

[NeurIPS 2025] Open-source Multi-agent Poster Generation from Papers

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/paper2poster--paper2poster/badge.svg)](/listings/paper2poster--paper2poster)

HTML

<a href="/listings/paper2poster--paper2poster"><img src="/api/listings/paper2poster--paper2poster/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Our Paper2Poster supports both local deployment (via [vLLM](https://docs.vllm.ai/en/v0.6.6/getting_started/installation.html)) or API-based access (e.g., GPT-4o).

**Python Environment**

pip install -r requirements.txt

**Install Libreoffice**

sudo apt install libreoffice

or, if you do **not** have sudo access, download `soffice` executable directly: https://www.libreoffice.org/download/download-libreoffice/, and add the executable directory to your `$PATH`.

**Install poppler**

conda install -c conda-forge poppler

**API Key**

Create a `.env` file in the project root and add your OpenAI API key:

OPENAI_API_KEY=<your_openai_api_key>

**Optional: Google Search API (for logo search)**

To use Google Custom Search for more reliable logo search, add these to your `.env` file:

GOOGLE_SEARCH_API_KEY=<your_google_search_api_key>
GOOGLE_SEARCH_ENGINE_ID=<your_search_engine_id>

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 279 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # 🎓Paper2Poster: Multimodal Poster Automation from Scientific Papers
  2. # 从学术论文自动生成学术海报
  3. <p align="center">
  4. <a href="https://arxiv.org/abs/2505.21497" target="_blank"><img src="https://img.shields.io/badge/arXiv-2505.21497-red"></a>
  5. <a href="https://paper2poster.github.io/" target="_blank"><img src="https://img.shields.io/badge/Project-Page-brightgreen"></a>
  6. <a href="https://huggingface.co/datasets/Paper2Poster/Paper2Poster" target="_blank"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Dataset-orange"></a>
  7. <a href="https://huggingface.co/papers/2505.21497" target="_blank"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Daily Papers-red"></a>
  8. <a href="https://x.com/_akhaliq/status/1927721150584390129" target="_blank"><img alt="X (formerly Twitter) URL" src="https://img.shields.io/twitter/url?url=https%3A%2F%2Fx.com%2F_akhaliq%2Fstatus%2F1927721150584390129"></a>
  9. <a href="https://huggingface.co/spaces/camel-ai/Paper2Poster" target="_blank"> <img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces%20Demo-blue"> </a>
  10. <a href="awesome-paper2poster.md" target="_blank"><img src="https://img.shields.io/badge/Awesome-Paper2Poster-brightgreen"></a>
  11. </p>
  12. We address **How to create a poster from a paper** and **How to evaluate poster.**
  13. ![Overview](./assets/overall.png)
  14. ## 📚 Awesome Paper2Poster
  15. New to this area? Check out **[Awesome Paper2Poster](awesome-paper2poster.md)** — a curated list of paper-to-poster methods, benchmarks, agent skills, and related paper-to-slides / video / webpage work.
  16. ## 🤩 Paper2Poster for Paper2Poster
  17. ![Overview](./assets/teaser.jpeg)
  18. ## 🔥 Update
  19. - [x] [2026.10.23] We release **[Awesome Paper2Poster](awesome-paper2poster.md)**, a curated list of paper-to-poster methods, benchmarks and agent skills.
  20. - [x] [2026.6.6] We introduce lightweight [Paper2Poster SKILL](https://github.com/Paper2Poster/Paper2Poster/tree/main/skills).
  21. - [x] [2025.10.7] Check out follow-up **[Paper2Video](https://github.com/showlab/Paper2Video)**.
  22. - [x] [2025.11.3] Added **Gradio demo** support.
  23. - [x] [2025.10.18] Added **Docker** support.
  24. - [x] [2025.10.13] Added automatic **logo support** for conferences and institutions, **YAML-based style customization**, a new default theme.
  25. - [x] [2025.9.18] Paper2Poster has been accepted to **NeurIPS 2025 Dataset and Benchmark Track**.
  26. - [x] [2025.9.3] We now support generate per section content in **parallel** for faster generation, by simply specifying `--max_workers`.
  27. - [x] [2025.5.27] We release the [arXiv](https://arxiv.org/abs/2505.21497), [code](https://github.com/Paper2Poster/Paper2Poster) and [`dataset`](https://huggingface.co/datasets/Paper2Poster/Paper2Poster)
  28. <!--## 📚 Introduction-->
  29. **PosterAgent** is a top-down, visual-in-the-loop multi-agent system from `paper.pdf` to **editable** `poster.pptx`.
  30. ![PosterAgent Overview](./assets/posteragent.png)
  31. <!--A Top-down, visual-in-the-loop, efficient multi-agent pipeline, which includes (a) Parser distills the paper into a structured asset library; the (b) Planner aligns text–visual pairs into a binary‐tree layout that preserves reading order and spatial balance; and the (c) Painter-Commentor loop refines each panel by executing rendering code and using VLM feedback to eliminate overflow and ensure alignment.-->
  32. <!--![Paper2Poster Overview](./assets/paperquiz.png)-->
  33. <!--**Paper2Poster:** A benchmark for paper to poster generation, paired with human generated poster, with a comprehensive evaluation suite, including metrics like **Visual Quality**, **Textual Coherence**, **VLM-as-Judge** and **PaperQuiz**. Notably, PaperQuiz is a novel evaluation which assume A Good poster should convey core paper content visually.-->
  34. ## 📋 Table of Contents
  35. <!--- [📚 Introduction](#-introduction)-->
  36. - [🌟 Awesome Paper2Poster](awesome-paper2poster.md)
  37. - [🧰 Agent SKILL](#-agent-skill)
  38. - [🛠️ Installation](#-installation)
  39. - [🐳 Docker Deployment](#-docker-deployment)
  40. - [🚀 Quick Start](#-quick-start)
  41. - [🔮 Evaluation](#-evaluation)
  42. ---
  43. ## 🧰 Agent SKILL
  44. We provide a lightweight skill (without the need of dependency) in `skills/` for AI agents such as Codex or Claude to help prepare poster-ready content from a paper.
  45. For example, in Codex, you can invoke:
  46. ```
  47. Use $paper2poster-poster to turn this paper into a poster package.
  48. ```
  49. See `skills/README.md` for more examples.
  50. ## 🛠️ Installation
  51. Our Paper2Poster supports both local deployment (via [vLLM](https://docs.vllm.ai/en/v0.6.6/getting_started/installation.html)) or API-based access (e.g., GPT-4o).
  52. **Python Environment**
  53. ```bash
  54. pip install -r requirements.txt
  55. ```
  56. **Install Libreoffice**
  57. ```bash
  58. sudo apt install libreoffice
  59. ```
  60. or, if you do **not** have sudo access, download `soffice` executable directly: https://www.libreoffice.org/download/download-libreoffice/, and add the executable directory to your `$PATH`.
  61. **Install poppler**
  62. ```bash
  63. conda install -c conda-forge poppler
  64. ```
  65. **API Key**
  66. Create a `.env` file in the project root and add your OpenAI API key:
  67. ```bash
  68. OPENAI_API_KEY=<your_openai_api_key>
  69. ```
  70. **Optional: Google Search API (for logo search)**
  71. To use Google Custom Search for more reliable logo search, add these to your `.env` file:
  72. ```bash
  73. GOOGLE_SEARCH_API_KEY=<your_google_search_api_key>
  74. GOOGLE_SEARCH_ENGINE_ID=<your_search_engine_id>
  75. ```
  76. ---
  77. ## 🐳 Docker Deployment
  78. For easier deployment, you can use Docker to run Paper2Poster without manual dependency installation.
  79. **Build the Docker image:**
  80. ```bash
  81. docker build -t paper2poster .
  82. ```
  83. **Troubleshooting:**
  84. - If you get a "permission denied" error when running Docker commands, use `sudo` before Docker commands (e.g., `sudo docker build -t paper2poster .`)
  85. **Example:**
  86. ```bash
  87. # Create output directory if it doesn't exist
  88. mkdir -p <4o_4o>_generated_posters
  89. docker run --rm \
  90. -e OPENAI_API_KEY=<your_openai_api_key> \
  91. -v "$(pwd)/Paper2Poster-data:/Paper2Poster-data" \
  92. -v "$(pwd)/<4o_4o>_generated_posters:/app/<4o_4o>_generated_posters" \
  93. paper2poster \
  94. python -m PosterAgent.new_pipeline \
  95. --poster_path="/Paper2Poster-data/<paper_name>/paper.pdf" \
  96. --model_name_t=4o \
  97. --model_name_v=4o \
  98. --poster_width_inches=48 \
  99. --poster_height_inches=36
  100. ```
  101. The generated poster will be saved in `<4o_4o>_generated_posters/Paper2Poster-data/paper_name/poster.pptx` on your host machine.
  102. ---
  103. ## 🚀 Quick Start
  104. Create a folder named `{paper_name}` under `{dataset_dir}`, and place your paper inside it as a PDF file named `paper.pdf`.
  105. ```
  106. 📁 {dataset_dir}/
  107. └── 📁 {paper_name}/
  108. └── 📄 paper.pdf
  109. ```
  110. To use open-source models, you need to first deploy them using [vLLM](https://docs.vllm.ai/en/v0.6.6/getting_started/installation.html), ensuring the port is correctly specified in the `get_agent_config()` function in [`utils/wei_utils.py`](utils/wei_utils.py).
  111. - [High Performance] Generate a poster with `GPT-4o`:
  112. ```bash
  113. python -m PosterAgent.new_pipeline \
  114. --poster_path="${dataset_dir}/${paper_name}/paper.pdf" \
  115. --model_name_t="4o" \ # LLM
  116. --model_name_v="4o" \ # VLM
  117. --poster_width_inches=48 \
  118. --poster_height_inches=36
  119. ```
  120. - [Economic] Generate a poster with `Qwen-2.5-7B-Instruct` and `GPT-4o`:
  121. ```bash
  122. python -m PosterAgent.new_pipeline \
  123. --poster_path="${dataset_dir}/${paper_name}/paper.pdf" \
  124. --model_name_t="vllm_qwen" \ # LLM
  125. --model_name_v="4o" \ # VLM
  126. --poster_width_inches=48 \
  127. --poster_height_inches=36 \
  128. --no_blank_detection # An option to disable blank detection
  129. ```
  130. - [Local] Generate a poster with `Qwen-2.5-7B-Instruct`:
  131. ```bash
  132. python -m PosterAgent.new_pipeline \
  133. --poster_path="${dataset_dir}/${paper_name}/paper.pdf" \
  134. --model_name_t="vllm_qwen" \ # LLM
  135. --model_name_v="vllm_qwen_vl" \ # VLM
  136. --poster_width_inches=48 \
  137. --poster_height_inches=36
  138. ```
  139. PosterAgent **supports flexible combination of LLM / VLM**, feel free to try other options, or customize your own settings in `get_agent_config()` in [`utils/wei_utils.py`](utils/wei_utils.py).
  140. ### Adding Logos to Posters
  141. You can automatically add institutional and conference logos to your posters:
  142. ```bash
  143. python -m PosterAgent.new_pipeline \
  144. --poster_path="${dataset_dir}/${paper_name}/paper.pdf" \
  145. --model_name_t="4o" \
  146. --model_name_v="4o" \
  147. --poster_width_inches=48 \
  148. --poster_height_inches=36 \
  149. --conference_venue="NeurIPS" # Automatically searches for conference logo
  150. ```
  151. **Logo Search Strategy:**
  152. 1. **Local search**: First checks the provided logo store (`logo_store/institutes/` and `logo_store/conferences/`)
  153. 2. **Web search**: If not found locally, performs online search
  154. - By default, uses DuckDuckGo (no API key required)
  155. - For more reliable results, use `--use_google_search` (requires `GOOGLE_SEARCH_API_KEY` and `GOOGLE_SEARCH_ENGINE_ID` in `.env`)
  156. You can also specify custom logo paths to skip auto-detection:
  157. ```bash
  158. --institution_logo_path="path/to/institution_logo.png" \
  159. --conference_logo_path="path/to/conference_logo.png"
  160. ```
  161. ### YAML Style Customization
  162. Customize poster appearance via YAML configuration files:
  163. - **Global defaults**: `config/poster.yaml` (applies to all posters)
  164. - **Per-poster override**: Place `poster.yaml` next to your `paper.pdf` for custom styling
  165. ## 🔮 Evaluation
  166. Download Paper2Poster evaluation dataset via:
  167. ```bash
  168. python -m PosterAgent.create_dataset
  169. ```
  170. In evaluation, papers are stored under a directory called `Paper2Poster-data`.
  171. To evaluate a generated poster with **PaperQuiz**:
  172. ```bash
  173. python -m Paper2Poster-eval.eval_poster_pipeline \
  174. --paper_name="${paper_name}" \
  175. --poster_method="${model_t}_${model_v}_generated_posters" \
  176. --metric=qa # PaperQuiz
  177. ```
  178. To evaluate a generated poster with **VLM-as-Judge**:
  179. ```bash
  180. python -m Paper2Poster-eval.eval_poster_pipeline \
  181. --paper_name="${paper_name}" \
  182. --poster_method="${model_t}_${model_v}_generated_posters" \
  183. --metric=judge # VLM-as-Judge
  184. ```
  185. To evaluate a generated poster with other statistical metrics (such as visual similarity, PPL, etc):
  186. ```bash
  187. python -m Paper2Poster-eval.eval_poster_pipeline \
  188. --paper_name="${paper_name}" \
  189. --poster_method="${model_t}_${model_v}_generated_posters" \
  190. --metric=stats # statistical measures
  191. ```
  192. If you want to create a PaperQuiz for your own paper:
  193. ```bash
  194. python -m Paper2Poster-eval.create_paper_questions \
  195. --paper_folder="Paper2Poster-data/${paper_name}"
  196. ```
  197. ## ❤ Acknowledgement
  198. We extend our gratitude to [🐫CAMEL](https://github.com/camel-ai/camel), [🦉OWL](https://github.com/camel-ai/owl), [Docling](https://github.com/docling-project/docling), [PPTAgent](https://github.com/icip-cas/PPTAgent) for providing their codebases.
  199. ## 📖 Citation
  200. Please kindly cite our paper if you find this project helpful.
  201. ```bibtex
  202. @misc{paper2poster,
  203. title={Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers},
  204. author={Wei Pang and Kevin Qinghong Lin and Xiangru Jian and Xi He and Philip Torr},
  205. year={2025},
  206. eprint={2505.21497},
  207. archivePrefix={arXiv},
  208. primaryClass={cs.CV},
  209. url={https://arxiv.org/abs/2505.21497},
  210. }
  211. ```

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.