Blog/Announcements
Announcements3 min read

Introducing Potato 2.0: AI-Powered Annotation

We're excited to announce Potato 2.0 with AI-powered features, multimedia support, and active learning capabilities.

By Potato Team·

Introducing Potato 2.0: AI-Powered Annotation

We're thrilled to announce the release of Potato 2.0, the most significant update to our annotation platform since its inception. This release brings AI-powered features, enhanced multimedia support, and sophisticated active learning capabilities that will transform how you create training data.

What's New in Potato 2.0

AI-Powered Annotation Assistance

Potato 2.0 introduces seamless integration with leading AI models including OpenAI GPT-4, Anthropic Claude, and Google Gemini. These integrations enable:

  • Smart Suggestions: AI models can provide annotation suggestions that annotators can accept, modify, or reject
  • Automatic Keyword Highlighting: Important terms and phrases are automatically highlighted to guide annotator attention
  • Quality Hints: Real-time feedback helps annotators understand potential issues with their annotations
# Enable AI assistance in your config
ai_support:
  endpoint_type: openai
  model: gpt-4
  ai_config:
    include:
      all: true

Enhanced Multimedia Support

Building on our text annotation foundation, Potato 2.0 adds robust support for:

  • Image Annotation: Classification, bounding boxes, polygons, and keypoint detection
  • Audio Annotation: Waveform visualization, transcription review, speaker diarization
  • Video Annotation: Frame-by-frame navigation, temporal event marking, object tracking

Active Learning Integration

Reduce your annotation effort by up to 50% with our new active learning module:

  • Sklearn Classifier Integration: Use any sklearn classifier to prioritize items where the model is least confident
  • Automatic Retraining: Models retrain as annotations accumulate, continuously improving sampling decisions
  • Flexible Configuration: Specify your classifier and feature extraction methods directly in YAML

Migration from Potato 1.x

Upgrading from Potato 1.x is straightforward. Your existing YAML configurations remain compatible, and we've added a migration tool to help you adopt new features:

# Upgrade your installation
pip install --upgrade potato-annotation
 
# Run the migration helper
potato migrate config.yaml --to-v2

Performance Improvements

Potato 2.0 is faster and more efficient:

  • 3x faster page loads through optimized rendering
  • 50% reduction in memory usage for large datasets
  • Real-time collaboration support for team annotation

Getting Started

Ready to try Potato 2.0? Installation is simple:

pip install potato-annotation
potato start your_config.yaml

Check out our Quick Start Guide for a complete walkthrough, or explore the Showcase to see example configurations.

What's Next

We're already working on Potato 2.1, which will bring:

  • Multi-modal annotation (annotate text, images, and audio together)
  • Enhanced admin dashboard with annotation analytics
  • Integration with Hugging Face datasets for seamless export

Thank you to our community of researchers and practitioners who have helped shape Potato. Your feedback drives our development, and we're excited to see what you build with 2.0.

Have questions or feedback? Join our GitHub Discussions or reach out on Twitter @PotatoAnnotation.