Thanks. That’s interesting, we are working on a similar crowd counting problem. Its fascinating to see how the models are getting a close ballpark estimate in an area that can be very tedious for us, humans.