Introduction

$$$$depth propto(focal length) * (real - world size)/(size in photo)

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

P * r == 1/Z(x) * exp(- expectation(lambda: (x, y)))

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

Z(x) == int(exp(- expectation(lambda: (x, y))) d y)

Introduction

$$$$depth propto(focal length) * (real - world size)/(size in photo)

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

P * r == 1/Z(x) * exp(- expectation(lambda: (x, y)))

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

Z(x) == int(exp(- expectation(lambda: (x, y))) d y)

Introduction

$$$$depth propto(focal length) * (real - world size)/(size in photo)

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

P * r == 1/Z(x) * exp(- expectation(lambda: (x, y)))

Our method

$$$$def expectation(l):

N=10000

return mean(l() for _ in range(N))

Z(x) == int(exp(- expectation(lambda: (x, y))) d y)