Image Recognition Instead of 2G Bar Codes?

This month’s Mobile Monday in Paris was quite a fruitful evening again. Ignacio Mondine of Daem Interactive gave an impressive presentation of a new technology they have developed which uses image recognition for mobile advertising and mobile commerce.

Instead of scanning a 2D bar code with a specialized application which leads to a web page via the phone’s browser, their technology is based on image recognition. Getting to the content behind an image is simple: The user takes a picture and sends it via MMS to the an image recognition server. The server then returns an SMS with a URL which can be accessed from the phone’s browser.

The main advantage of this approach compared to 2D bar codes is that no special software has to be installed on the phone. In addition, it looks like there are a number of different companies out there working on 2D bar code solutions and their bar codes are not compatible with each other. If several of them get themselves established, several bar code readers on the phone might be required. I can hardly imagine that this will be accepted by users. So from this point of view, image recognition might even be the natural next step after 2D bar code solutions and the market (except for Japan I guess) might even jump completely over 2D bar codes right into image recognition solutions which do not require extra software on the handset.

A disadvantage of the image recognition approach might be that images can not be recognized because of the image quality. Especially in low light situations, pictures tend to be grainy and blurry. When taking pictures from magazine ads, users also tend to position the phone too close to the image and the picture will not be sharp. A possible solution to this is also already in sight. Instead of taking a picture which is then sent via MMS, Deam Interactive is already taking the next step and are developing a system which uses 3G video calls connected to an image recognition server in the network. Lots of possibilities here, including augmented reality applications about which I have written before.

While still being a small startup company, they’ve already managed to get a couple of operators behind them for some trials. This is crucial for any kind of mobile marketing as the user can only receive content free of charge if the operator bills the advertiser and also for mobile commerce as micro payment is also very difficult to do without the mobile operator in the middle.

3 thoughts on “Image Recognition Instead of 2G Bar Codes?”

  1. I agree that installing a client is a major hindrance. On the other hand, sending pics to a server to analyze them without being sure to get a result sounds to me even more dangerous. And it is costly too for both parties. And how do you bill the trials and errors?

    The good thing about the cool barcode readers is that the code is analyzed and resolved on the phone and only then a server connection happens.

  2. Couple of other disadvantages to add to that already mentioned.

    Sending images can be slow and expensive. Typically, you’re talking a few hundred kilobytes on a 2MP built in camera. And that will grow when phones have higher resolution cameras. Yes, you can scale down the images taken but, in the real world, most people will not bother or necessarily know how to do that. Fiddling with settings is something most people aren’t willing to do in an ad-hoc basis. Compare that with 2 or so kilobytes of data needed to transfer 2D codes I think the advantage is clear.

    Yes, it’s a pain to have to install a 2D reader and that will put some people off but then again it’s a once off process and there’s always the possibility that either the handset manufacturer or the network provider will install it on the phone, providing an out of the box solution. I think what’s holding them back is that there’s not yet a standard for 2D code that everyone is happy to adopt. QR codes are the closest and very popular in Japan but it’s not a worldwide standard and there are drawbacks to it.

    The second drawback is that there’s not enough of a strong ‘call to action’ which you have with 2D codes. When you see a 2D code, it draws you attention and you’d know what to do with it. I would argue that not having a distinct call to action, is a clear disadvantage. Yes you can support the image with text instructions but that’s self defeating if the idea is to capture the casual browser/viewer.

Comments are closed.