blend_non_transparent multiplication error

[rhthomas]

The program runs until the webcam on my laptop is engaged, then it fails with the following error. Looks like a matrix size mismatch.

...
Neural Network loaded!
Traceback (most recent call last):
  File "camera.py", line 19, in <module>
    output = sudoku_master(img)
  File "songoku/image_parsing.py", line 40, in sudoku_master
    img_final = blend_non_transparent(img_original, img_sudoku_final)
  File "songoku/helpers.py", line 95, in blend_non_transparent
    face_part = (face_img * (1 / 255.0)) * (background_mask * (1 / 255.0))
ValueError: operands could not be broadcast together with shapes (720,1280,3) (480,640,3) 

[rhthomas]

Ah so the webcam on-board my device is 1280x720, so one needs to modify the perspective_transform function:

def perspective_transform(img, transformation_matrix, original_shape=None):
    warped = img

    if original_shape is not None:
        if original_shape[0] > 0 and original_shape[1] > 0:
            warped = cv2.resize(
                warped,
                (original_shape[1], original_shape[0]),
                interpolation=cv2.INTER_CUBIC,
            )

    white_image = np.zeros((640, 480, 3), np.uint8)

    white_image[:, :, :] = 255

    # warped = cv2.warpPerspective(warped, transformation_matrix, (640, 480), borderMode=cv2.BORDER_TRANSPARENT)
    warped = cv2.warpPerspective(warped, transformation_matrix, (640, 480))

    return warped

Swapping 640 for 1280, and 480 for 720 in this function stops the program crashing. Could be worth adding functionality to detect the dimensions of the camera at start-up.