Questions for the seminar Paper "Tracking Everything Everywhere All at Once"
-----------------------------------------------------------------------------------------------
Please send your answers to: argusm@informatik.uni-freiburg.de


1) What is taken as input for the optimization process?

2) What two networks are learned during optimization, what are their inputs and outputs? How do image frame features flow into the prediction?

3) Which quantities are aggregated via blending, and which terms is the loss composed of?