Welcome to Eydara
Clarity at the edge of your production line.
What is Eydara?
Section titled “What is Eydara?”Eydara is an on-premise platform for visual quality inspection. It connects cameras on a production line to AI models that help detect product defects — and puts domain experts in control of what those models learn.
The platform is built around a pipeline with six stages. Each stage handles a distinct part of the journey from raw image to deployed model.
The Eydara pipeline
Section titled “The Eydara pipeline”| Stage | What it does |
|---|---|
| Capture | Cameras at inspection rigs photograph products as they move along the production line. |
| Preprocess | Raw images are adjusted — corrected, cropped, or enhanced — to prepare them for analysis. |
| Ingest | Processed images are stored and registered so they can be accessed for review and annotation. |
| Annotate | Domain experts review images and label what they see. This is where EyVz works. |
| Train | Annotations are used to train AI models that learn to recognize defects from your labels. |
| Deploy | Trained models are deployed back to the production line, where they begin assisting with inspection. |
Where EyVz fits
Section titled “Where EyVz fits”EyVz is the tool you use during the Annotate stage. It is where you — the person who knows the product — inspect captured images, review what the AI model predicts, and provide your own labels.
The annotations you create in EyVz are what the models learn from. When a model is uncertain or wrong, your corrections teach it to do better next time.
What you can do in EyVz
Section titled “What you can do in EyVz”- Browse and filter images captured from your production line
- Annotate images — label defects, confirm predictions, or correct mistakes
- Review model predictions — see what the AI thinks, and how confident it is
- Set confidence thresholds — decide when a prediction is strong enough to trust
- Compare model versions — evaluate whether a newly trained model performs better than the current one
Next steps
Section titled “Next steps” Home view Browse and filter inspection images.
Annotate from scratch Start labeling when no model exists yet.
Understand model behavior Evaluate what a model gets right and wrong.
Glossary Definitions for terms used in EyVz.