CODEWITHSUNDEEP
numpy
somebody4
somebody4

Reputation: 515

Numpy, filter rows use mask of 1,0 values?

Opencv sometimes return a mask for filtering.

Give array A=[[1,2],[3,4],[5,6]]

and mask mask=[1,0,1]

How should I apply the mask to obtain [[1,2],[5,6]]?

I tried A[mask==1] but it says dimension not match.

np.where and np.nonzero seem not working too.

Edit:

Turns out A[mask==1] works,
It is in the real case I faced that mask.shape is (n,1) but not (n,)
that extra 1 caused the trouble.
np.squeeze solved the problem

Upvotes: 2

Views: 938

Answers (2)

AMC
AMC

Reputation: 2702

Both 

mask = mask.nonzero()
res = A[mask]

and 

mask = mask.astype(bool)
res = A[mask]

should work!

Upvotes: 1

Tinu
Tinu

Reputation: 2503

Like this?

A = np.array([[1,2],[3,4],[5,6]])
mask = np.array[1,0,1])
>>> A[np.where(mask==1),:]
array([[[1, 2],
        [5, 6]]])

Upvotes: 1

Related Questions