Supported factory-owned cameras
Reuse a suitable existing camera or choose a supported new one. Image quality, optics, lighting, timing and interfaces are qualified against the inspection decision.
Qualify camera readinessHow it works
From production images to a working line inspection, in days.
United Vision manages the no-code inspection workflow from model training in your chosen, customer-owned cloud account to deployment on your supported edge devices. Use your own qualified cameras, move a suitable visual inspection into production in days, then update it as products, processes and defects change.
In brief
AI visual inspection learns to recognize a defined condition from representative production images. United Vision manages training in the cloud account your company chooses and owns, then deploys the approved model to a supported factory-owned edge device. Production inference always runs at the edge, not in a vendor-owned SaaS training or inspection environment.
Factory-controlled architecture
Your cameras. Your edge. Your preferred cloud.
United Vision connects your qualified cameras, your chosen cloud account and your supported edge devices in one no-code model workflow. Training runs in the customer-owned cloud account, not a United Vision-owned SaaS account. Approved models run at the edge, so a production verdict does not wait for a cloud round trip.
Your chosen cloud account
United Vision manages training and retraining in a supported cloud account your company chooses and owns. Your production images, trained AI models and customer-specific inspection IP remain yours.
Understand data ownershipYour factory-owned edge
The platform deploys the approved model to your supported edge device, where production inference always runs. Qualify compute capacity and camera-to-response timing for the actual line.
Assess edge readinessShorten the path to the first inspection, the next use case and the next update.
First inspection: qualify the image and response, then train and validate for production. Next use case: repeat the same platform workflow under the factory license. Production change: add new evidence, retrain in your cloud and deploy the approved update to your edge.
The lifecycle
From first inspection to continuous improvement, in ten steps.
Define
State the condition, decision and response.
Capture
Collect representative production images.
Label
Apply plant knowledge to regions and classes.
Train
Train candidate models in your cloud account, without writing AI code.
Validate
Test against agreed acceptance criteria.
Deploy
Deploy the approved model to your supported factory-owned edge.
Operate
Run production inference at the edge and trigger the configured response.
Review
Inspect errors, uncertainty and change.
Retrain
Add new evidence and retrain in your customer-owned cloud account.
Redeploy
Validate, approve and deploy the updated model to the edge.
Quality gate
Release with evidence your plant can trust.
Reserve evidence the model did not train on.
Validation should reflect the products, conditions and edge cases the workflow is expected to encounter.
Connect model output to process risk.
False accepts, false rejects and uncertain results have different consequences and need an agreed response.
Know what triggers review.
New products, materials, suppliers, lighting, tooling, camera position or defect definitions can require revalidation.
Questions answered
What buyers ask about how it works.
How long does an inspection take to deploy?
United Vision is designed to deploy suitable visual inspection use cases in days. The actual schedule depends on image availability, camera and lighting readiness, line integration and the validation required for the production decision.
Do we need developers or data scientists to train and deploy inspections?
No. United Vision provides a no-code workflow for supported model training, deployment and updates. People who know the product define and label the relevant conditions; the platform manages the AI workflow. Camera setup, lighting, line integration and production validation still need appropriate engineering and implementation support.
What happens when production changes?
Use the same workflow to review new evidence, update defect classes and retrain in your chosen, customer-owned cloud account. Validate the candidate model against current production conditions, then deploy the approved update to your supported edge device. The platform manages these model steps without requiring the team to write AI code.
Who owns the cameras, edge device and cloud account?
The customer owns and chooses the supported cameras, edge devices and cloud account. United Vision manages model training in that customer-owned cloud account, not in a vendor-owned SaaS training environment, and deploys models to the factory edge. Camera, optics, lighting, edge capacity and interfaces are qualified for each inspection.
Does a line verdict require a cloud round trip?
No. Production inference always runs on the supported edge device at the factory. Training and retraining run in the cloud account your company chooses and owns. The cloud supports the model lifecycle, but is not the real-time camera-to-verdict path.
Can the same platform support more than one inspection?
Yes. One recurring factory software license supports unlimited United Vision use cases within the licensed factory. Repeat the same no-code training, deployment and update workflow for the next inspection without another United Vision software purchase. Each use case still needs suitable evidence, imaging, integration and production validation.
Bring one inspection opportunity. Leave with a clearer path.
A working session is designed around the image source, operating condition, decision, response and ownership model, not a generic product tour.