Hi, I have the same problem.
My ‘image’ was created using camera = cv2.VideoCapture(0); return_value, img = camera.read()
Have you moved further with this issue?
Open ‘site-packages/gluoncv/data/transforms/presets/ssd.py’ file, so if your gluoncv is latest, you will see the code as below in ‘transform_test’ method.
if isinstance(imgs, mx.nd.NDArray):
imgs = [imgs]
for im in imgs:
assert isinstance(im, mx.nd.NDArray), "Expect NDArray, got {}".format(type(im))
Check these commands: ‘isinstance((your array), mx.nd.NDArray)’ and ‘type((your array))’
If you get ‘false’ and ‘mxnet.ndarray.ndarray.NDArray’, I think it is a bug.