Find rotation angle of binary image in opencv python

Question

I have a set of images containing top views of solar panels as seen bellow :

In order for me to get the rotation angle of the solar panels,I trained a model to segment the images as seen bellow:

In order to extract the angle of rotations I used cv2.findContours then fitline, fitellipse, minareareact but the angle various were not consistent

 area = cv2.contourArea(contours[i])
        if area>max_area:
            cnt = contours[i]
            max_area = area
            
    rect = cv2.minAreaRect(cnt)
    x, y = [], []
    ((centx0,centy0), (widthe0,heighte0), angle_200)= cv2.fitEllipse(cnt)

I also used PCAcompute butthe angles zere notconsistent either

Now, I am following another approach in which I apply canny edge detection to extract the edges and then extract the angles:

Or this one

But even here using the different fitting techniques did not turn out to be robust.

Is there a good method to extract angles of rotations or tilt of the panels from this images?