INDUSTRIAL VISION

Industrial vision & manufacturing

Turn subtle differences into useful inspection data.

Organize image and video data for surface defects, component recognition and operational workflows. Align capture conditions and defect boundaries with clear inspection rules.

Discuss your project Explore applications

Gear assembly, fixture and components beneath an inspection camera — INDUSTRIAL VISIONFrom real-world signals to structured data.

APPLICATIONS

Start with the task your model needs to learn.

Connect each application to its data, processing rules and deliverables.

THE TASK / 01

Surface defect recognition

Define scratches, dents and stains; distinguish visible defects from normal texture and inconclusive samples.

WHAT THE DATA CONTAINSDefect classes, region contours, occlusion and uncertainty flags

See inputs, labels and formats

THE TASK / 02

Components & assembly state

Organize components by station and viewpoint. Define missing-part, misalignment and occlusion rules, retaining normal samples.

WHAT THE DATA CONTAINSComponent boxes, state attributes and image-to-station indexes

See inputs, labels and formats

THE TASK / 03

Operations & events

Segment defined operations into events, associate object tracks, and record interruptions, occlusion and out-of-frame periods.

WHAT THE DATA CONTAINSEvent clips, temporal boundaries and object associations

See inputs, labels and formats

DATA IN · VALUE OUT

A clear brief. A traceable delivery.

A representative sample helps establish scope before batch work begins. Delivery fields and acceptance criteria are agreed during the pilot.

Missing data?Plan sources and a collection pilot Already have data?Check files, indexes and pairings first

01 / WHAT TO SHARE: Task goals, product categories, capture conditions, defect definitions and shareable representative samples.
02 / WHAT YOU RECEIVE: Raw-data index, label files, class dictionary, rule version, review records and issue list.
03 / WHAT TO REVIEW: Review by defect class and scenario; agree how to handle boundary decisions, missing labels and incorrect labels.

QUALITY IN CONTEXT

Align defect boundaries before scaling batches.

Capture conditions

Record devices, stations, lighting and materials; check blur, glare and field of view.

Rule alignment

Agree the rules using normal, defective and borderline samples; send uncertain items for review.

Batch separation

Group by product, device or production batch to avoid near-duplicate leakage across splits.

Explore delivery & quality

PROJECT PERSPECTIVE

See how a task becomes a data project.

Four metal surfaces showing normal texture, a scratch, a dent and a stain

MFG-001

From surface variation to consistent defect boundaries

Use normal textures and defect samples to agree annotation rules for metal surfaces. Deliver region labels with indexes that support review.

Read case study

BEFORE WE START

Common questions

01Can we start with few defect samples?

Start by auditing available samples, defining classes and identifying gaps, then run a small collection or annotation pilot. Limited samples do not justify model-performance promises.

02How are rules aligned across production lines?

Use a shared class dictionary and record device, material and station differences. Resolve ambiguous boundaries with reference examples and pilot review.

LET’S BUILD YOUR DATA

Tell us about your task and data.

Start with your task, the data available and one representative example. We can then discuss a practical pilot scope.

Discuss your project All industry applications