Nice paper. But is it really "seemingly implausible" that you can predict "what you are doing" from "how you move" (even if that is just reconstructed from camera poses)?

Amazing, Richard! Very happy for you!