Voyance is a computer vision data platform that helps teams produce high quality ground truth from image datasets with speed, control, and auditability, including bounding boxes and semantic segmentation.
Voyance treats the ontology as the source of truth for what labels mean and what is allowed, so dataset definitions stay consistent instead of drifting over time.
At the core is an interactive ontology feedback loop that lets you explore your data, edit and restructure categories, and ground the ontology in evidence from the dataset itself, reducing QA churn and keeping labeling decisions explicit and user guided.
Your ontology defines what labels mean and drives consistency across every annotator and dataset revision.
Explore, restructure, and refine categories in real time through a live feedback loop grounded in your actual data.
Every labeling decision is tracked and explainable so you always know why a category changed.
Go from raw images to production ready ground truth faster with intelligent suggestions and bulk operations.
Have questions or want to learn more? We'd love to hear from you.