Setting up WineEngine

Once you have everything you need ready for WineEngine, you or your developers should first review the WineEngine API documentation.

Once you or your developers are familiar with your intended integration of WineEngine, you can sign up and get started.

Signing up for WineEngine

You can sign up for WineEngine here. Choose the plan that best fits the amount of images in your label database, as well as how many searches you estimate your users will perform. It’s often best to start small with the Starter plan so you can scale your plan with your usage.

Once you’ve entered your details and signed up for access to your API, your request is processed by our engineers at TinEye, who will create a private server for your WineEngine API, customized for your usage.

Adding your label images to WineEngine

Once you have access to your WineEngine API server, the first thing you should do is add your images. WineEngine supports images up to 1000 pixels on the smaller side for fingerprinting. If your images are larger than that we recommend that you resize them before submitting them to WineEngine for indexing, because smaller images will make the uploads faster.

Images are added one at a time using the add method. Images can be uploaded directly to the server or you can specify a URL where the image is located and the WineEngine server will download it from there.

The WineEngine API can perform one add operation at a time. Any extra requests submitted will be queued up for processing as slots become available. While adding images one at a time might seem slow, in practice it is possible to load over 100,000 images per day. So even very large collections can be added in a reasonable amount of time.

Interpreting your results

Once your images have been added, you’re ready to perform searches. Submitting a search call to your API server will get a response with matching images from your collection — including their score and match_percent.

The score is the relative quality of the match; it specifies how closely the two images match.

The match_percent indicates the size of the matching regions of the two images as a percentage. If this is lower than 10%, it will indicate that a very small region of the two images matches together: this is a good threshold for you or your developers to use to eliminate false positives.

Adding metadata to WineEngine

Once WineEngine has matched a photograph with one of the product labels in your image collection, you’ll want to access additional data about that product, such as prices, ratings or reviews. By design, WineEngine doesn’t have the ability to store metadata, it’s intended to work with your own database. When you’re adding images to your collection, set the filepath to your database’s internal ID for that product. That filepath will be included in search results, which you can then use to get the metadata from your own systems.