So we released the second version of our API earlier today. Great, but what does this actually mean?

New Interface

If you thought our API was complicated, give it another look. We have let go of the legacy XML-based format and opted for JSON. The documentation has also been reworked in a way that should let you write your first application in a matter of minutes. Just check out how ridiculously easy it is to write a full-featured client in Bash (with cURL).

If you have been using using API v1 in your application, do not worry: we have maintained compatibility for you. New users, however, must use the new interface.

New Features

Synchronous Indexing

Previously, the service to index an image was asynchronous. That meant that there was a delay between the time when you did so and the time when you could scan it. Not anymore! New images can now be looked for instantaneously.

Reference Images Upload

As a consequence of the above we no longer require you to host your reference images on the Web: you can now send them directly from your desktop machine or any other client. You can even integrate this service in a mobile application! Allowing your users to create new search-able objects has just become much easier.

Statistics as a Service

Since the first release of our API we have been providing usage statistics on our extranet. Now you can also access those statistics programatically. This should be much appreciated by advanced users who want to integrate them in proprietary dashboards as well as developers (like me) who live in a command prompt.

New Engine

This is another important change that you will probably not notice, but this API update was simultaneous with an upgrade of our image search engine. It comes with near realtime writes and increased scalability.

I don’t want to get into too much detail now because I know you are eager to try the new API, so here is the tl;dr version of how it works:

  • • a collection of services written in Ruby;
  • • bindings to our home-made image recognition libraries;
  • • processes communicating through MessagePack-RPC;
  • • index completely stored in memory thanks to Redis.

Free Tier

All the above got you curious about our API? Don’t hesitate to try by yourself: it’s free up to reasonable limits.