Why Setting XRayResizer engine to cv2 could increase performance.

[LaskawiecPiotr]

I think there is a small error in the XRayResizer.

Currently the code is:

Xray Resizer class XRayResizer(object):
    """Resize an image to a specific size"""

    def __init__(self, size: int, engine="skimage"):
        self.size = size
        self.engine = engine
        if 'cv2' in sys.modules:
            print("Setting XRayResizer engine to cv2 could increase performance.")

    def __call__(self, img: np.ndarray) -> np.ndarray:
        if self.engine == "skimage":
            with warnings.catch_warnings():
                warnings.simplefilter("ignore")
                return skimage.transform.resize(img, (1, self.size, self.size), mode='constant', preserve_range=True).astype(np.float32)
        elif self.engine == "cv2":
            import cv2  # pip install opencv-python
            return cv2.resize(img[0, :, :],
                              (self.size, self.size),
                              interpolation=cv2.INTER_AREA
                              ).reshape(1, self.size, self.size).astype(np.float32)
        else:
            raise Exception("Unknown engine, Must be skimage (default) or cv2.")

Printing the message makes no sense if one set engine="cv2" Wouldnt it make more sense to have it as

        if 'cv2' in sys.modules and self.engine != "cv2":
            print("Setting XRayResizer engine to cv2 could increase performance.")

Sorry, if this is too trivial to mention.

[ieee8023]

Good catch. I removed the message all together: #165