Skip to main content

Discover

Youzu Lens: visual search for e-commerce.

Youzu Lens lets shoppers search your catalog with an upload, an image URL or an object detected in a photo; matches can be exact, similar or broad. It is visual search for e-commerce that works from your catalog's own product images, so products need images.

Key points

  • Use a photo, an image URL or an object detected in a photo to search your catalog.
  • See exact, similar or broad matches from your own product images.
  • Use the selected object when a photo holds several; the results follow it.
  • Keep your existing text search.

You choose how strict a photo search is: exact, similar or broad. Results carry no score.

Try it

Search with a photo.

Choose a sample, select the item that matters and search.

Image URL

https://example.com/navy-sofa.jpg

A navy velvet sofa with one long seat cushion and wooden legs.
Try with a sample

Selected item: Whole photo

4 results from the example catalog

  • A navy velvet sofa with one long seat cushion and wooden legs.

    Velvet 3-seater, navy

    Exact

    Sofa · Velvet · Navy · 3 seats

  • A cream linen 3-seat sofa with a wooden base and soft cushions.

    Linen 3-seater

    Similar

    Sofa · Linen · Cream · 3 seats

  • A cream 3-seat sofa with purple, red, pink and grey cushions and a taupe throw.

    Cream 3-seater

    Similar

    Sofa · Cream · 3 seats

  • 2 plum velvet seats side by side, seen from the side, with dark legs.

    Velvet loveseat, plum

    Broad

    Sofa · Velvet · Plum · 2 seats

Read as text
  1. Choose a sample, select the item that matters and search.
  2. Pick an object in a photo: a generated room photo labeled Virtually staged, with 4 detected items: Sofa, Accent chair, Coffee table, Floor lamp. The sofa is selected first.
  3. Sofa: Linen 3-seater (Similar); Cream 3-seater (Similar); Velvet 3-seater, navy (Broad); Velvet loveseat, plum (Broad).
  4. Accent chair: Accent chair, plum (Similar); Armchair, terracotta (Broad); Velvet loveseat, plum (Broad).
  5. Coffee table: Oak coffee table (Similar); Oak side table (Broad).
  6. Floor lamp: We could not find a useful match in this assortment. Try a closer image or select a different item.
  7. Upload a photo: Armchair, terracotta (Exact); Accent chair, plum (Similar); Velvet loveseat, plum (Broad).
  8. Paste an image URL (Image URL: https://example.com/navy-sofa.jpg): Velvet 3-seater, navy (Exact); Linen 3-seater (Similar); Cream 3-seater (Similar); Velvet loveseat, plum (Broad).
See it with your products

Decisions

How a decision is made

Inputs
An upload, an image URL or an object detected in a photo, and your catalog's own product images.
Signals
Visual similarity between the query and your product images. Each result carries a match label of exact, similar or broad.
Match labels
No review step: results are shown to the shopper. Your storefront decides how many to show and which page each result opens.
Your catalog
Lens works from your catalog's product images and returns results; your storefront decides what happens next.
Uploaded images
Ask us how uploaded images are retained and deleted.

The example's first state: the room photo with the sofa selected.

The example's first state: the room photo with the sofa selected.
QueryResultMatch label
Photo of a room, sofa selectedLinen 3-seaterSimilar
Cream 3-seaterSimilar
Velvet 3-seater, navyBroad
Velvet loveseat, plumBroad

Try it

Find it from a photo

Choose an object in the photo, predict how close the catalog's top match is, then see the catalog re-order.

Virtually staged: an entryway with a light oak slatted bench, a black dome pendant lamp and a seagrass basket on the floor.Virtually staged
Objects in the photo
Predict the closest match

Choose an object first.

Example catalog

  • A low upholstered oatmeal bench on thin black metal legs.
    Upholstered bench
  • A grey felt storage bin with tan leather handles.
    Felt storage bin
  • A matte black pendant lamp with a wide shallow shade.
    Black saucer pendant
  • A light oak slatted bench with a lower open shelf.
    Oak bench with shelf
  • An empty pine slatted storage crate.
    Pine crate
  • A round white paper lantern pendant.
    Paper lantern
  • A light oak slatted bench on slim splayed legs, without a shelf.
    Oak bench, splayed legs
  • A black metal wire basket with 2 handles.
    Wire basket
  • A brushed brass dome pendant lamp.
    Brass dome pendant
  • A sand-coloured linen storage box with a lid.
    Linen storage box
  • A dark walnut slatted bench with a lower open shelf.
    Walnut bench with shelf
  • A clear ribbed glass globe pendant lamp.
    Ribbed glass pendant
Read as text

The photo shows an entryway with 3 objects: 1 Bench, 2 Pendant lamp, 3 Basket. The example catalog has 12 products. For each object you predict the closest match: the same product, a similar product or something broader. Then the catalog re-orders and the top match shows its label.

  • Bench: top match Oak bench with shelf, exact. The catalog has this same oak bench with the shoe shelf, so the top match is exact. Next: Walnut bench with shelf (similar), Oak bench, splayed legs (similar), Upholstered bench (broad).
  • Pendant lamp: top match Brass dome pendant, similar. The catalog has this dome lamp in brass only, so the top match is similar. Next: Black saucer pendant (similar), Paper lantern (broad), Ribbed glass pendant (broad).
  • Basket: top match Felt storage bin, broad. The catalog has no woven basket, so the top match is broad: another storage piece. Next: Linen storage box (broad), Pine crate (broad), Wire basket (broad).

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.

Visual search works from your catalog's own product images; products need images.

See it with your products

See it in the product

Find catalog products from an image.

Pick an object in a picture, and Youzu Lens returns the matching products from your catalog.

Questions buyers ask

Youzu Lens is visual search for e-commerce. A shopper searches your catalog with a photo, an image URL or an object detected in a photo, and the matches can be exact, similar or broad. It works from your catalog's own product images, so products need images. Try the example above with the sofa photo.

An uploaded photo, an image URL, or an object detected in a photo. When a photo holds several items, the shopper selects the one that matters and the results follow that selection. Search starts from an image upload or an image URL. Read the Lens documentation for the exact inputs.

No. Lens returns candidates from your catalog, each with a match label of exact, similar or broad. An exact match needs the same product to be in your assortment with usable images; otherwise the best results are similar or broad matches, and your interface should say which is which. Test your own assortment in a demo.

No. Text search stays in your existing search, and Lens adds a photo route beside it. You decide where the camera or upload entry appears and which product page each result opens. Youzu also adds cleaner product data where your attributes are thin. The Discover page explains where the placements fit.

Youzu does not publish a search time. Latency depends on your catalog, images and integration, so Youzu measures it on your catalog during the evaluation. The film and the example on this page show how matches are presented; they are not a performance measurement. Ask for the measurement on your own data.

Supported categories are scoped with you, and Youzu does not claim Lens works for every category. Results depend on the quality and coverage of your product images. Bring a sample of your assortment, including cluttered scenes and items you do not stock, and review the results together in a demo.

Next step

See it with your products.

Bring the photos your shoppers use.