Now you do not have to get bored of searching for a particular picture if you can just search it by just typing down the details about. Even if the uploader has not provided caption to the picture or the caption is just a imaginative pictorial representation and is not related to the characters present in it Google will get those images for you.
Because Google has already succeeded in generating caption of an image depending on the characters present in it. If you want a dog with a hat just type it down and you will get those pictures fitting to your needs.
Because Google has already succeeded in generating caption of an image depending on the characters present in it. If you want a dog with a hat just type it down and you will get those pictures fitting to your needs.
You might be imagining how can this be done but it has already done successfully by Google Search. Following are the answers to your curiosity:
Pic1
We have already aware face recognition that we usually use in our cameras and mobiles. Google have evolved this technology by using Deep Learning methodology. Goolge implemented it for not only recognizing objects in a pictures but also classifying those objects in a single picture(eg., different type of colored balls on the billiard table). After this it made the Google search to narrate down the description of the image.
Pic2
First need of auto captioning a picture is to get the picture data that is done by vision algorithms and then it is merged with natural language processing to produce a description. This may sound easy and simple but they are quite complex. And this has been done by combining recent computer visions and language models into single trained system by using two Recurrent Neural Networks.
No comments:
Post a Comment