---
title: README
emoji: 👀
colorFrom: green
colorTo: green
sdk: static
pinned: false
license: mit
thumbnail: >-
  https://cdn-uploads.huggingface.co/production/uploads/65d65a40531e0bc924f0b1a3/6tV18h0u5JX0YtrZreK4E.png
---
<div style="display: flex; align-items: top; justify-content: space-between;">
  <div>
    <h1>PictSure Model Family</h1>
    <ul>
      <li><strong>Repo:</strong> <a href="https://github.com/PictSure/pictsure-library">PictSure on GitHub</a></li>
      <li><strong>Paper:</strong> <a href="https://www.alphaxiv.org/abs/2506.14842">arXiv:2506.14842</a></li>
      <li><strong>Website:</strong> <a href="https://pictsure.github.io/">PictSure.eu</a></li>
    </ul>
    <h3>🔍 Summary</h3>
    <p>
      PictSure is a lightweight vision-only in-context learning (ICL) family of models for few-shot image classification (FSIC).
      It uses pretrained frozen embeddings and outperforms CLIP-based models on out-of-domain tasks like medical imaging.
    </p>
    <p>
      <strong>Contacts:</strong>
      <a href="mailto:lukas.schiesser@dfki.de">lukas.schiesser@dfki.de</a> & 
      <a href="mailto:cornelius.wolff@cwi.nl">cornelius.wolff@cwi.nl</a>
    </p>
  </div>
  <div>
      <img src="https://cdn-uploads.huggingface.co/production/uploads/65d65a40531e0bc924f0b1a3/6tV18h0u5JX0YtrZreK4E.png" alt="PictSure Logo" width="600"/>
    </div>
</div>