Is there a way to convert a tensor object to an array so as to work with OpenCV functions?

Question

I am working with an image dataset where I want to preprocess images using some OpenCV functions. To load the dataset I\'m using the following code:

train_ds = tf