CODEWITHSUNDEEP
pythonnumpyscikit-learnscikit-image
James Owers
James Owers

Reputation: 8345

Creating a label dataset from a segmented image using Python

I've labeled an image to produce a numpy array with labels e.g.

array([[0, 1, 0, ..., 0, 0, 0],
        [0, 1, 0, ..., 0, 0, 0],
        [0, 1, 0, ..., 0, 0, 0],
        ..., 
        [0, 0, 0, ..., 0, 0, 0],
        [2, 2, 0, ..., 0, 0, 0],
        [2, 2, 0, ..., 0, 0, 0]], dtype=uint8)}

what is the most efficient way to turn this into the dataset:

x-coord | y-coord | label 
-------------------------
0       | 0       | 0
0       | 1       | 1
0       | 2       | 0
...
1024    | 0       | 2
1024    | 1       | 2

etc.

I don't mind what the output format is but I expect a dictionary would be most convenient.

This is my current slow code that iterates through coordinates of the image:

(x, y) = img.shape
for x1, x2 in np.ndindex((x, y)):
    data[(x1, x2)] = img[x1, x2]

The reason I'm doing this is that I would like to add other features to an array for each pixel.

Upvotes: 0

Views: 1028

Answers (1)

Divakar
Divakar

Reputation: 221594

You could use np.meshgrid and np.vstack to create a Nx3 numpy array having a similar format as stated as desired in a vectorized manner, like so -

In [103]: img
Out[103]: 
array([[0, 1, 1, 0, 0],
       [0, 1, 0, 0, 1],
       [1, 1, 1, 1, 2],
       [2, 1, 1, 0, 2]])

In [104]: M,N = img.shape

In [105]: Y,X = np.meshgrid(np.arange(N),np.arange(M))

In [106]: np.vstack((X,Y,img)).reshape(3,-1).T
Out[106]: 
array([[0, 0, 0],
       [0, 1, 1],
       [0, 2, 1],
       [0, 3, 0],
       [0, 4, 0],
       [1, 0, 0],
       [1, 1, 1],
       [1, 2, 0],
       [1, 3, 0],
       [1, 4, 1],
       [2, 0, 1],
       [2, 1, 1],
       [2, 2, 1],
       [2, 3, 1],
       [2, 4, 2],
       [3, 0, 2],
       [3, 1, 1],
       [3, 2, 1],
       [3, 3, 0],
       [3, 4, 2]])

Upvotes: 1

Related Questions