Paul Zhou
Community@Paul-Jeo · Shanghai
I am Paul-Zhou, a postgraduate student at SHU. My research direction is AI4SSCI.
Agent Skills by Paul Zhou
Showing 1 vetted skills indexed across 1 GitHub repositories.
Frequently Asked Questions About Paul Zhou
FAQPage SchemaWhat tasks can I accomplish with Paul Zhou's Image2PPT skill?▼
Image2PPT converts slide images, screenshots, scanned PDFs, and image-only PPT/PPTX files into high-fidelity object-level editable PowerPoint decks. It performs semantic-region mixed reconstruction, rebuilds flowcharts, knowledge graphs, composite shapes, and arrows, and preserves speaker notes. It is not intended for authoring new decks from notes.
Who is the target user for the Image2PPT skill?▼
The skill targets students, researchers, educators, and office professionals who need to recover editable presentations from static images, screenshots, or scanned documents. It suits anyone restoring legacy image-based slides into fully editable PowerPoint objects rather than creating presentations from scratch.
How does Image2PPT handle diagrams and complex graphics?▼
Image2PPT performs measured reconstruction of flowcharts and knowledge graphs, producing native circle nodes, connectors, and single-object thin or filled arrows. Composite graphics are rebuilt as editable PowerPoint objects, and a rendered QA step verifies the fidelity of the reconstructed output.
What inputs does Image2PPT accept and what are its limits?▼
Accepted inputs include slide images, screenshots, scanned PDFs, and image-only PPT/PPTX files, covering Chinese use cases like 图片转可编辑PPT and 扫描PDF恢复. The skill explicitly does not author new decks from notes; it only reconstructs existing visual slide content into editable form.