SKUVis

IMAGE TOOL

Element recognition for product photos

Audit packaging creative before you spend credits on erase or image translation.

At a glance: Upload a product image and select element types to detect on the product and/or background; SKUVis returns a recognition report (text, boxes, optional stats) for 1 credit—files outside 256–3000 px per edge or 10 MB are rejected before quoting.

Bounding boxes on product packaging elements
Detection report in the workspace—then chain smart erase or image translation if needed.

What you get

A structured recognition report—not automatic translation or erasure.

Choose detect types separately for elements on the product vs the background: watermark, logo, text, text block. Optional flags return recognized characters, text share, product count/share, or border pixels when enabled in UI.

Element types

You must pick at least one type on the product or background or SKUVis blocks submit.

Use the report to decide whether smart erase, image translation, or manual retouching is the next step.

  • Watermark
  • Logo
  • Text
  • Text block (colored regions with copy)

How it works in SKUVis

Upload, toggle detect categories and optional report fields, confirm 1 credit, read results in the workspace.

  1. Upload within 256–3000 px per edge and 10 MB.
  2. Select element types on product and/or background.
  3. Toggle optional report fields (recognized text, proportions, etc.).
  4. Confirm quote, review report, open erase or translation tools if needed.

Recognition does not modify pixels—it only analyzes the uploaded image.

What it is not

Not legal label review, not guaranteed OCR accuracy on stylized fonts.

Have a human verify packaging claims before publishing in a new market.

Inputs & outputs

What you provide

  • One product image
  • objectDetectElements / nonObjectDetectElements (types 1–4)
  • Optional report flags (text, proportions, counts)

What you get

  • Recognition report in workspace / History
  • 256–3000 px per edge
  • 10 MB max
  • 1 credit—confirm in app

Common use cases

  • Scope erase before supplier cleanup
  • List visible text before image translation
  • QA intake batches from factories

FAQ

Does recognition translate automatically?

No—open image translation when you want localized on-image copy.

Same types as smart erase?

Erase uses removal categories; recognition uses detect categories (watermark, logo, text, text block) scoped to product vs background.

If recognition fails, are credits returned?

Failed or canceled runs show as not charged in History when release rules apply.