Image recognition — how far along are we?

March 14, 2018

Usually when people think of image recognition or mixing photo and VR their first thought goes to functionalities such as unlocking your phone with facial recognition or animojis or filters used for entertainment in smartphones and social media.

Image recognition is the study of certain patterns between multiple images, using artificial intelligence processes, in order to label the content and then use it for specific purposes.

There is already a lot of software that focuses on photo recognition and is able to recognize text, age ranges, brands and even emotions.

It is possible for users to train some models, for instance, IBM Watson, and there are models such as Google Cloud Vision API that are trained every time a user searches for images.

This kind of technology is already being applied in a series of events to maintain security but the possibilities are endless.

Police in China are already using this technology in glasses with image recognition capabilities that can spot drivers and license plates, in order determine whether a specific car has been stolen, whether the person driving the car has a criminal record, etc…

There are many useful applications and it is easy to train the models to some extent, since everyone nowadays posts ever more photos on social media.

This kind of technology may be used to improve security in many ways, by recognizing possible suspects, license plates, cars, etc… but it may also be dangerous when used for bad purposes.

There is currently no strict set of regulations covering what is an experimental technology that is still being developed, meaning that some companies may already have a lot of facial expressions and data available for use. This is dangerous because a security breach can occur and a lot of facial data could be in the wrong hands.

This kind of information can be regulated and a user may opt to stop sharing their data; the main problem is that many users are not aware that they are sharing it.

As we have discussed before, photo recognition is already being used in marketing processes, as it is extremely useful to target ads and provide better customer support.

Brands can easily be spotted in customers’ photos and not only can patterns be studied such as “this product is commonly used in the kitchen” or “customers usually take pictures with our product when by the ocean”, but this also makes possible to provide improved customer support. By reviewing various social media posts where a specific brand is present, companies can, for example, easily understand when customers are not happy by facial expressions, and then proceed to contact that person and understand the reasons why.

Facial recognition may be a useful technology when carefully used because, as with fingerprints, each person is unique and it adds an additional security layer when dealing with sensitive data and personal content.

If you like what you read, please take a moment to share it with others:

Start growing your business now.