Hi kuanfang, thank you for your excellent work. I found that the "resize_points" function seems to be confusing width and height:
|
def _resize_points(points, src_shape, dst_shape): |
|
"""Resize the points.""" |
|
xs = points[:, 0] |
|
ys = points[:, 1] |
|
new_xs = xs * dst_shape[0] / src_shape[0] |
|
new_ys = ys * dst_shape[1] / src_shape[1] |
Here shape[0], [1] should be width, height, respectively. But according here,
|
image_shape = (int(height), int(width)) |
shape[0], [1] is height, width, respectively.
Could I trouble you to clarify which one is correct, thank you!
Hi kuanfang, thank you for your excellent work. I found that the "resize_points" function seems to be confusing width and height:
opra/utils/dataset.py
Lines 17 to 22 in b4649e3
Here shape[0], [1] should be width, height, respectively. But according here,
opra/utils/dataset.py
Line 81 in b4649e3
shape[0], [1] is height, width, respectively.
Could I trouble you to clarify which one is correct, thank you!