Suppose that we want to segment a red blob from the image, normally you will have a class for this red blob e.g. 0. And every red blob you detected will have a class of 0.

But, in my case, I want that the model will look at the surrounding context, e.g., if the red blob is surrounded by blue blobs, it should be classified as class 1, instead of 0. Like the following image.

enter image description here

Is this something easily achieve able with U-Net or other models (you can suggest)?

In my case, the context can be more difficult than this, e.g., if there are blue and green surrounding you, you will have another class.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.