Skip to main content

Glossary

What is a match level?

A match level states how close a visual search result must be to the query image: exact, similar or broad.

In plain words: A visual search returns products that look like a photo. The match level says how close that is: exact for the same item, similar for the same kind of item with differences, broad for a looser likeness. It lets a shop show the closeness instead of one undifferentiated list.

Published by Youzu · Editorial policy

Published

Why it matters

Shoppers use a photo in different ways. One wants the very item; another wants something like it.

How it works

  1. Compare

    Measure how close each catalog image is to the query image.

  2. Label

    Place each result at the exact, similar or broad level by how close it is.

  3. Show

    Display the level with the result, and order the list with the closest first.

One query, 3 result rows
LevelWhat it showsShopper reading
ExactThe same itemThis is the one
SimilarThe same kind, with a differenceClose, but check the detail
BroadA looser likenessSomething in the same spirit

Examples

  • A photographed sofa

    An upload of a navy velvet sofa returns the same sofa as exact.

  • A jacket on a street

    A detected olive jacket in a street photo matches the catalog jacket at the exact level when the cut and color agree.

  • A chair with no close match

    A worn chair has no exact match in the catalog, so only similar and broad results are shown.

Common mistakes

  • One undifferentiated list. Mixing exact and loose matches in one list makes a shopper guess which result is the item they photographed.

  • Broad shown first. Putting looser results above closer ones makes the best result harder to find.

Visual search
Finding products from an image instead of a text query, by comparing the image's visual features with those of catalog items.
Product matching
Deciding whether 2 records describe the same product by comparing identifiers, titles, attributes and images.
False merge
A false merge joins 2 records that describe different products; a missed match leaves 2 records of the same product apart.

In Youzu

Youzu helps marketplaces, classifieds and retailers turn raw seller and supplier product data into structured, reviewable catalog records.

Youzu Lens

Sources

  1. Content-based image retrieval at the end of the early years, IEEE Transactions on Pattern Analysis and Machine Intelligence 22(12). doi.org/10.1109/34.895972. Accessed .

  2. Vision API Product Search documentation, Google Cloud. docs.cloud.google.com/vision/product-search/docs. Accessed .

FAQ

Questions buyers ask

Youzu Lens lets shoppers search your catalog with a photo: an upload, an image URL or an object detected in a photo; matches can be exact, similar or broad. The search runs against the products already in your catalog, so a level describes closeness to your own items.

Yes. Visual search works from your catalog's own product images; products need images. A product with no usable image has nothing to compare with the shopper's photo, so it cannot appear in a visual result until you add one for it.

Not as a rule. A photo answers a question that words describe badly, such as a look or a shape, while a typed query suits a known name or code. Many shops keep both and let shoppers pick the one that fits the moment.

Last reviewed

Next step

Automate Commerce. Make shopping personal.

Bring your catalog, listing flow or shopping idea. Let's find the right place to start.

Book a demo

Tell us about your catalog and we will get back to you.