Getting Started¶
Welcome to the Annotation Garden Initiative (AGI). This guide will help you get started with our tools for collaborative neuroscience stimulus annotation.
Why AGI?¶
The Annotation Gap Problem¶
Naturalistic stimuli datasets (movies, complex images, audio) contain rich, multi-dimensional information ideal for neuroscience research. However, these datasets typically arrive with minimal or no annotation, despite having diverse dimensions suitable for analysis.
This creates a fundamental barrier: while methodologies like hyperalignment and functional connectivity can operate without annotations, most analytical approaches (GLM-based analytics, encoding models) rely heavily on them.
The result? Despite the wealth of neural data collected with naturalistic stimuli, researchers often default to resting-state paradigms simply because the annotation burden is too high.
A New Science Paradigm¶
AGI represents a paradigm shift: instead of each lab independently annotating the same stimuli, we build shared, versioned, community-curated annotation layers.
Rich annotations unlock:
- Novel analytics on complex neural data
- Cross-study comparisons using common stimuli
- FAIR-compliant datasets ready for re-analysis
- Reduced barrier to entry for naturalistic neuroscience
Projects like StudyForrest and Neuroscout have demonstrated the value of rich annotations; AGI provides the infrastructure to scale this approach.
Quick Start Paths¶
Choose your path based on what you want to do:
-
Generate HED Annotations
Use HEDit to convert natural language descriptions into valid HED annotations.
-
Annotate Images
Use VLMs to automatically annotate neuroscience stimulus images.
-
Understand Standards
Learn about HED, BIDS, and Stim-BIDS integration.
-
Contribute Annotations
Add your annotations to shared stimulus repositories.
HEDit Quick Start¶
1. Install HEDit¶
2. Get an API Key¶
HEDit uses OpenRouter for LLM access. Get your API key and initialize:
3. Generate Annotations¶
# From text description
hedit annotate "A red circle appears on the left side of the screen"
# From an image
hedit annotate-image stimulus.png
# Validate existing HED
hedit validate "Sensory-event, Visual-presentation"
4. Use the API¶
import httpx
response = httpx.post(
"https://api.annotation.garden/hedit/annotate",
json={"description": "A face image is shown for 500ms"},
headers={"X-OpenRouter-Key": "your-key"}
)
print(response.json()["annotation"])
Image Annotation Quick Start¶
1. Clone the Repository¶
2. Set Up Environment¶
3. Run the Service¶
4. View Dashboard¶
Visit http://localhost:3000 to browse annotations.
Next Steps¶
- Installation Guide: Detailed setup instructions
- HEDit Documentation: Full HEDit reference
- Standards: Learn about HED and BIDS
- Contributing: Add your annotations