Use own images set to create voxelized shape

[quintendewilde]

How can I use a set of own images to create a new (no example based) voxel shape.

I want to use pots (images I have from them to make a 3D voxel shape)

The instruction show only the learning aspect and not how to use the code on new images. Or I'm not fully understanding the shapenet part.

UPDATE

Changed the demo.py code a bit and put some pictures in the imgs directory.

When running python demo.py newobject.obj

I got this error.

Traceback (most recent call last):
  File "demo.py", line 84, in <module>
    main()
  File "demo.py", line 52, in main
    demo_imgs = load_demo_images()
  File "demo.py", line 43, in load_demo_images
    return np.array(ims)
ValueError: could not broadcast input array from shape (3,300,300) into shape (3)

I have no idea what this means.

Tried resizing the images to a smaller width and length but that still gives me the same error.

[Metal-joker]

same issue

[quintendewilde]

Anything changed? Anybody an idea to train own images??

[Metal-joker]

Still no solution

[heromanba]

The images of training dataset have a shape of (127, 127, 3). So you need to resize your image to this shape so that the network can test the output using you image.

[kitachan]

Convert your images to size 127 x 127.

[ramyagopinathpsg]

I converted the image size to 127 x 127 but still its not working

[ramyagopinathpsg]

Convert your images to size 127 x 127.

its not working when i changed the image size to 127 x 127