When you call mypolygon == point, numpy performs element wise comparison maintaining dimensions of the coordinates:

• it checks for -1 in the first column (x-coordinates)
• it checks for 0 in the second column (y-coordinates)

np.where() returns array with indices by dimension, where the values in point occur in mypolygon (at same columns). It is "3-dimensionnal" because it found 3 occurences:

• -1 at one index in mypolygon (1st column): at [2, 0]
• 0 at two indexes in mypolygon (2nd column): at [0, 1], [8, 1]

for what you are trying to achieve, you could use:

print(np.all(mypolygon == point, axis=1))

• np.all(..., axis=1) checks if all coordinates matches along each row
• True indexes will get you the n-th row values that matches point