I am working on my first multi label classification problem and have run into a couple of questions in regards to location of object in image.
The first thing I am wondering about is using
randomResizeCrop on an image with multiple labels and objects. For example say the crop cuts out one of the labeled objects, wont it still be training with the original group of labels as though some part of the removed object is still showing?
The second thing I am looking for is a way to visualize bounding boxes and labels over the objects in the image. In the multi classification example from the book it shows color filters over each object in the image but I have had a hard time finding info regarding this
any relevant info appreciated!