Auto Tagging (Auto Tagging only available in the Pro Version)

It is common knowledge that auto tagging without any human input is not possible. Therefore how have we been able to achieve it with about 99,99999% certainty.  We used our own version of neural networking and a certain amount of learnt knowledge. The idea is our idea and concept alone and is unique to us and we own all intellectual property rights to the following idea now explained.

When you start using the General Picture Recognition Software after completing a Search using the Picture Recognition Engine it returns image that are the same, or close to the sample image. These images are tagged so that they can be easily found in the future. 

What auto tagging does is use those tagged images (we know they are 100% correct because they are tagged by a human) and learns importantly automatically from those already tagged images.  The neural network treats them as learnt and correct neurons, it then compares them with images that are not the same but very close then it tags automatically those new untagged images. 

Because most similar images are stored in the same folder its default is to only use already tagged images in a folder and use them to tag other images in the same folder.  However there are options that allow you to use all tagged images to tag images in all  folders. Auto tagging will go through the whole of your hard drive tagging images automatically and will not even need to be checked.  But if you wish to check the results if only to say WOW then they can be listed. 

The default options are set so that certainty is just about 99.999999%.  The software user can lift this certainty to 100% but this will limit the amount of files found.  But also its true of the opposite you can select options that reduce certainty but still using auto tagging.  If certainty is reduced then images should be checked and manually tagged if required.  By reducing certainty the software will jump to less similar images but usually the same main object will be found. The organizer can be used to re-tag if required.  Auto tagging is iterative it then uses those new learnt tagged images to tag other untagged images.

On this image you notice that the software is automatically tagging the image. The software is looking at all the already tagged images and comparing them with an untagged image.  We use a neural network type approach to achieve this and provide options to increase or degrease how close the matches need to be before an image is tagged. Now in the latest version we use scenarios to scan for images to be tagged in the background while working on other things.

General Picture Recognition Software