Hello, thanks for your great work and patient replyment.
I find that the inputs of the proposed nn surfemb are cropped images where there're only one instance, whenever do training or inference, aren't they? So can you deal with a image which has multiple instances whose bboxs are not known?
Hello, thanks for your great work and patient replyment. I find that the inputs of the proposed nn
surfembare cropped images where there're only one instance, whenever do training or inference, aren't they? So can you deal with a image which has multiple instances whose bboxs are not known?