Object detection without bounding boxes

by Ambesh   Last Updated September 11, 2019 09:19 AM

I'm trying to think of an architecture to perform object detection from labelled images (according to object class) but any information in the labels with respect to object location. I thought I could setup a sliding window and a CNN, but I'm wondering how to train this model as the sliding window won't find the object most of the times. Is there any type of loss function that can handle this?

Thanks in advance!



Related Questions


Updated March 13, 2019 05:19 AM

Updated May 09, 2019 17:19 PM

Updated September 11, 2017 20:19 PM

Updated April 10, 2018 21:19 PM

Updated May 01, 2019 15:19 PM