Yes – there are three settings to consider in making a private visual search system:

(1) Training a private database
You can upload your own set of training labeled images using the Training API (http://developer.iqengines.com/apidoc/current/apis/training-api.html). The default setting is that we match your query images against both your own database and the IQ Engines public image database. However, if you would like to search your images exclusively, we can set your database to “private”, so when a photo with your API key arrives, we only search your private database, and no one else has access to those learned objects/images.

(2) Turn off crowdsourcing
By default, if we do not detect a match between your query image and an image from your database, then the image is sent to human crowdsourcing engine for labeling. However, if your application does not necessitate crowdsourcing (your objects of interest are such that computer vision can recognize them), we can turn off crowdsourcing, and only perform computer vision search in your private database. Please note – IQ Engines also can make a “private” crowdsourcing node that specializes in your application/location. Contact us if you want to learn more about private crowdsourcing nodes.

(3) Real-time mobile client search
If your application is such that you are trying to recognize only a few hundred objects, you can perform the search on the mobile device (not using the server). This way, your visual query searches your private dataset and is not sent to the visual search server. This real-time local mobile client search is a feature that IQ Engines will offer soon as part of our image intelligence platform – stay tuned!