Image Detection vs. Emitters

A colleague, in the process of planning a mobile experience, asked us whether he should consider using image detection or Bluetooth emitters (aka beacons) to serve up content associated with a specific outdoor location. He had some beautifully designed markers and image detection seemed like a good way to go. We’ve conducted some specific technical research into both emitters and image detection (might share some of that later) but here are our big picture, general thoughts, on the subject:

Image Recognition and Its Limitations

In our experience, and in general, emitters are way better than image detection. The problem with image recognition is two fold:

1. It isn’t good enough

Image recognition technology is actually pretty weak and requires a ton of processing power from the device. This is because image detection is really hard for the software on the device. In order for the device to recognize the object it needs to capture and interpret the image. When the image is incredibly simple like a QR code it can usually handle it. Even then the device usually requires that the QR code be captured at a specific angle, distance, etc. This forces the user to position the device and the image exactly in a frame. Again, a device can usually handle a QR code, a more complex image is likely beyond the capability of current image recognition technology to perform reliably.

Most of the augmented reality apps you see don’t actually use image recognition but actually use GPS and the accelerometer to roughly overlay content. It isn’t great and usually the documentation far outstrips the experience.

2. It’s unnatural

By forcing the user to place an object in frame you are actually asking a lot. People are lazy and asking them to pull out, lift and position a device is usually more than a user is willing to go unless the reward is spectacular. This is why people don’t use QR codes. It’s too much effort with too little reward.

Why emitters are better

Emitters function similar to EZPass at highway tolls. It’s completely passive and transparent. You can essentially keep on driving. The environment gives the device cues in a format it can understand and doesn’t need to exert any effort to process. The user can keep the device in a position that is comfortable for them and isn’t asked to do very much to interact with the experience. All in all it’s much more magical and transparent.

That said, emitters aren’t perfect and have their limitations as well (perhaps more on this soon) for example they have spotty support on Android devices.

So, here are two outdoor emitters we like:

Image credit: Estimote