Getting error "'NoneType' object has no attribute 'astype'"

Python
1

I have been trying to implement a project on POINT CLOUD COMPLETION but have been facing the error ‘NoneType’ object has not attribute ‘astype’, I have checked the path of the files which is correct and I have installed opencv 3.4.1 on python2.7 because this is an old project and requires old libraries.
I am getting error in add_kinect_v1_noise of pc_cmpletion_util file which is getting called from the main file in load_image “im=add_kinect_v1_noise(im)”
And it shows error on “dm_m = dm.astype(float) * 0.0001 / scale / 1000” in pc_completion_util
The main file of the project is this :

def load_image(self, file_name, add_noise = False):
im = read_png16(file_name)
if add_noise:
im = add_kinect_v1_noise(im)
im = depthmap_16bitTofloat(im)
return im
This is the func from main file which calling the func add_kinect_v1_noise from util files.

def read_png16(file_name):
CV_LOAD_IMAGE_ANYDEPTH = 2 # apparently leftover from cv before cv2
int16_d_image = cv2.imread(file_name, CV_LOAD_IMAGE_ANYDEPTH)
return int16_d_image

def add_kinect_v1_noise(dm): # operates on the uint16
scale = 0.003
dm_m = dm.astype(float) * 0.0001 / scale / 1000

p_world = dm2points(dm_m)
X = np.reshape(p_world[0,:], (128,128))
Y = np.reshape(p_world[1,:], (128,128))
Z = np.reshape(p_world[2,:], (128,128))

if np.min(Z) <= 0.01:
    print ("ZERO?")

nz, angles = surfnorm_fast(X,Y,Z)

dm_m_noise = add_noise(dm_m, Z, angles)

return (dm_m_noise * 1000 * scale / 0.0001).astype(np.uint16)
2

Hi,
you want us to debug 700 lines?

May be you can try to use a debugger and set a breakpoint before error to investigate…

3

No I am asking to debug all the lines I have just given all the code for reference, I’ll edit the post so that only the part with issue is left.

4

this must be the 5162th “imread() returned an empty image, and python user did not expect that” post.

1 Like
5

But the path is correct I have printed the path to see if it is correct or not

6

go and check the output from imread(), please

7

you check that using os.path.exists(), in case you were wondering.