When I read a colour image in OpenCV, it is showing the dimensions as 256x256x3. But I need to pass it as 3x256x256 array to my neural network. How do I change the array shape, retaining the pixel locations in BGR.
1 Answer
You can simply transpose the array. For an example, my picture is a 10 x 10 picture:
import numpy as np
#my picture
wrong_format = np.arange(300).reshape(10,10,3)
correct_format = wrong_format.T
If it works properly, then correct_format(0,1,1) should be equal to wrong_format(1,1,0). And we can see that it is:
correct_format(0,1,1) == wrong_format(1,1,0)
True
Comments
