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.
Virtually stagedOpen a sample: Pick an object in a photo · Upload a photo · Paste an image URL
Press Search to see matches from the example catalog.
Read as text
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.
| Query | Result | Match label |
|---|---|---|
| Photo of a room, sofa selected | Linen 3-seater | Similar |
| Cream 3-seater | Similar | |
| Velvet 3-seater, navy | Broad | |
| Velvet loveseat, plum | Broad |
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.

Objects in the photo: 1 Bench, 2 Pendant lamp, 3 Basket.
Example catalog
Upholstered bench
Felt storage bin
Black saucer pendant
Oak bench with shelf
Pine crate
Paper lantern
Oak bench, splayed legs
Wire basket
Brass dome pendant
Linen storage box
Walnut bench with shelf
Ribbed glass pendant
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.
Documentation
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.