Select a region
Microsoft has launched a new app that can analyse and caption photos using artificial intelligence to identify what’s in the image.
The software, called CaptionBot can be used by anyone. It tries to learn and understand what is in a photo before offering up an opinion as to what is in it, with varying degrees of accuracy.
Microsoft is quick to point out that the software is still very new, and the more people test it out, the better it will become at identifying images.
The software is being driven by two separate pieces of technology. The first is image recognition, which recognises different features in an image to help identify it. The second is natural language, which uses human descriptions of objects and pairs them with features in order to produce captions that sound human-like.
The software can be quite hit-and-miss, sometimes getting the subject matter spot on, other times having a few more issues. For early stage technology, though, it is nonetheless impressive and follows in the footsteps of other Microsoft image tech from their research team – including programs that aim to guess age and emotions shown in photographs. Google has created similar algorithms in the past too.
