I am working on a problem statement that how to correctly identify the same objects which is varying in size or structure. For example: in the image there are a number of chairs but each of which is of different shape and size (Like few are small chairs and few are big likewise).
Would you give your opinion on how to deal with this?
You can train a neural network. Your training dataset shall contain objects of the same category but with various structures and sizes. For example, different kinds of chairs. You can also use data augmentation changing scale to add size variation. Check out Localizer, this is a python library for object detection with varying sizes, structures, and orientations.
Hi! I would recommend to take a look at Inception neural network.
It uses different sizes of convolution kernels to take variations into account. This way you don't need to expect the object to be properly positioned or to occupy a certain area.