Summary

This repository offers an AI-powered toolkit for reproducing and generating academic paper figures, targeting researchers and power users. It aims to significantly reduce repetitive plotting tasks by providing systematized style parameters derived from high-impact conference papers, enabling both data-driven generation and image-based replication.

How It Works

The toolkit provides two core skills: plot-from-data and plot-from-image. plot-from-data leverages pre-defined, systematic style parameters extracted from nine real paper figures, allowing users to input their data according to specific academic aesthetics. plot-from-image enables users to upload a figure screenshot, which the system analyzes for proportions, fonts, and colors to automatically generate a corresponding matplotlib script for reproduction.

Quick Start & Requirements

Setup involves utilizing Python scripts within the repository structure (e.g., plot-from-data/scripts/, plot-from-image/scripts/). Primary dependencies are Python and matplotlib; specific version requirements or hardware constraints (GPU, CUDA) are not detailed. Users can find parameter documentation and example scripts within the repository. For user-submitted examples and discussions, refer to https://github.com/Trae1ounG/paper-plot-skills/issues/1.

Highlighted Details

• Features a curated gallery of nine distinct plot styles, including paired bar charts with delta arrows (MemEvolve), grouped bar charts with hatch fills (SPICE), line plots with confidence bands and inset zooms (Self-Distillation, SiameseNorm), t-SNE scatter plots with clustering (MemGen), broken-axis scatter plots (Meta-Harness), and dual-series radar charts (DoRA).
• The plot-from-image skill supports automatic reproduction of complex layouts, such as table-like class-wise results, directly from uploaded screenshots.
• Styles are meticulously extracted from top-tier conference papers, preserving specific aesthetic elements like font types (serif, sans-serif, LaTeX), line styles, fill patterns, and axis configurations.

Maintenance & Community

The repository is presented as a starting point and is subject to continuous optimization based on user reproduction processes. There are no explicit mentions of core maintainers, sponsorships, or dedicated community channels like Discord or Slack. Users are encouraged to star the repository to support ongoing development.

Licensing & Compatibility

The specific open-source license for this repository is not stated in the provided README content. Consequently, compatibility for commercial use or linking within closed-source projects cannot be determined without further clarification.

Limitations & Caveats

The project is described as a "starting point" and undergoing "continuous optimization," suggesting it may be in an evolving or beta phase. The accuracy and robustness of the plot-from-image feature across a wide variety of input image qualities and plot complexities are not detailed. No known bugs or unsupported platforms are listed.