PaperBanana: Automating Academic Illustration for AI Scientists

arXiv 2026 49.4 method, system, benchmark, application

TLDR

PaperBanana automates publication-ready academic illustration generation using VLMs and image models, with a benchmark from NeurIPS 2025.

Reasoning

The paper addresses a practical bottleneck in research workflows with a well-defined framework and benchmark. Its strength lies in empirical evaluation and domain-specific focus, but it is limited to illustration generation rather than broader scientific discovery.

Read-first score

Read-first score 49.4, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 39.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,benchmark,evaluation,experiment

Topical relevance 42%

32.5

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 51.

Keyword Scores

AI for scientific research

AI scientist

paper writing agent

research automation

autonomous research agent

automated research

automated scientific discovery

literature review agent

scientific discovery agent

survey generation

automated experimentation

experiment design agent

Deep Analysis

Innovations

Agentic framework orchestrating specialized agents for reference retrieval, content/style planning, rendering, and iterative self-critique refinement
PaperBananaBench, a benchmark of 292 methodology diagrams from NeurIPS 2025 for evaluating academic illustration generation

Methodology

PaperBanana uses an agentic framework with VLMs and image generation models, where agents retrieve references, plan content and style, render images, and iteratively refine via self-critique. Evaluation is performed on PaperBananaBench, a curated set of 292 methodology diagrams from NeurIPS 2025, comparing against leading baselines on faithfulness, conciseness, readability, and aesthetics.

Key Results

PaperBanana consistently outperforms leading baselines across faithfulness, conciseness, readability, and aesthetics, and also generates high-quality statistical plots.

Resources

Paper