and if you want to try image recognition for yourself, you could have a go with the blippar app, whichas well as faces. for certain types of objects, it has been trained on specially curated datasets, categories like cars, landmarks, works of art, dogs and cats and flowers, which means that once it has recognise an object it can go into quite a bit of detail about it. and also show you its knowledge graph which is a web of all be connected themes that surround the object that it can see. blippar is really made for augmented reality marketing campaigns and also educational experiences, which means it's better at recognising some things than others. and, let's face it, it is trying to do two really hard tasks, here: first pick out an object in the scene, and then somehow match it to its training images. but it can recognise more general objects under the right conditions. the things it finds hardest to recognise are objects which are deformable, floppy, and don't keep their shape, like clothing and politicians‘ opinions. and yes, it can also do famous faces. it‘s trawled the web to pull