Library
I want to annotate an image automatically with Belief Networks. Until now I have extracted texture and color features from images. But the problem is how to design a Belief Network.
Quote 0 0
Library
I’m speculating you already have a set of images coming with their annotations, and you want to learn a BBN to automatically tag new images with your annotations. If yes, this is a typical supervised learning application. Therefore, you have to prepare a dataset with as many lines as already annotated images, and as many columns as images’ characteristics (texture and color features). The last column will contain the images’ annotations.You just have to define the last column as your Target variable, and use one of the available BayesiaLab supervised learning algorithms (Naïve, Augmented Naïve, Augmented Markov Blanket, …).supervisedLeearning.jpeg Once the BBN learned and validated, you would be able to use it for automatically associating an annotation to new images, given their features, either through interactive or batch inference. You would also be able to use the BayesiaLab's APIs to integrate this tagging process into your own program.
Quote 0 0