Web App · Visualization · Annotation
Neuroscout web tools
Browser tools for naturalistic movie research in the Neuroscout ecosystem: a p5.js feature visualizer for high-dimensional stimulus annotations, plus a web platform we used to label audiovisual movies for downstream neuroimaging, behavior, and ML.
- Stack
- JavaScript · p5.js · web platform
- Ecosystem
- Neuroscout · naturalistic stimuli
- Outputs
- Explorable features · structured labels
Overview
This entry combines two related pieces of lab infrastructure for the same workflow: understanding and annotating naturalistic movies for neuroscience.
The feature visualizer is an interactive p5.js sketch that plots feature time series from Neuroscout / AudioSet-style extractors so researchers can see what the model is tracking moment by moment—turning opaque vectors into something inspectable.
Alongside that, I built and deployed a web-based annotation platform for manual labeling of movie stimuli: flexible task flows, structured exports for modeling, and collaboration across labs. Same motivation as the visualizer—high-dimensional, messy stimulus data needs tooling people can actually use.
Feature visualizer
- Visualizes feature time series aligned to naturalistic stimuli
- Supports feature sets available through Neuroscout
- Connects extraction pipelines to human-interpretable events during a film
Annotation platform
- Flexible labeling workflows for different studies
- Structured outputs for neuroimaging, behavioral, and ML pipelines
- Deployment and integration with ongoing projects and collaborators
Why it matters
Both pieces reflect the same habit: research infrastructure that makes messy inputs debuggable and reusable—whether you are exploring features visually or producing labels for the next analysis stage.
My role
Designed and built the p5.js visualizer and its integration with feature tables; designed, implemented, and deployed the annotation platform and supported its use in downstream workflows.