After about three weeks of working tensorflow API for object detection, I am now able to see bonuding box around some test images. It is quite an exciting experience for me!

My question is:

when the model is trained using tensflow Records,( which has h,w,x,y ) along with the other features of the images, do h,w,x,y also go thru neural network process, i,e, does the forward and backward propagation apply to bounding box parameters as well.

Thanks in Advance