>>13130707bit rusty on my stats, but would need to know first if you want exactly 90%, at least 90%, or going with a bit of a margin at least 80%.
assuming your general population is big enough to be "infinite enough" so you can ignore when taking any individual out of the population, and let's make obesity rate 50% for ease of calculation and checking of the calculation.
this means each woman standing there had 50% chance of being fat.
now, let's round to 10 women and let's go with your number of 90% (asian at the very end is clearly not fat, and there are a few which are chubby/fat and not obese, but whatever).
let's try and get exactly 90% (which is near enough at least 90% in this case as well)
10 coinflips, need to get 9 heads (which, considering their size, should be easier than getting tails).
Odds are about 1 in 100.
doing at least 80% obese drops it down to about 1 in 20.
the odds should be worse as we're using 50%, and not 40%, but it gives a ballpark figure
