State of the Art.
Cham and Rehg ‘99
•
Single camera, multiple hypotheses.
•
2D templates (solves drift but is view dependent)
I(
x
,
t
) = I(
x
+
u
,
0) +
h
