Caffe: classify.py skimage error: ValueError: can not convert object to float64

[Kenosuke]

Hi, I receive this error when running the python wrapper classify.py on ubuntu 12.04 server

Traceback (most recent call last):
  File "classify.py", line 161, in <module>
    main(sys.argv)
  File "classify.py", line 113, in main
    inputs = [caffe.io.load_image(str(args.input_file))]
  File "/home/ubuntu/caffe/caffe-dev/python/caffe/io.py", line 18, in load_image
    img = skimage.img_as_float(skimage.io.imread(filename)).astype(np.float32)
  File "/usr/local/lib/python2.7/dist-packages/skimage/util/dtype.py", line 287, in img_as_float
    return convert(image, np.float64, force_copy)
  File "/usr/local/lib/python2.7/dist-packages/skimage/util/dtype.py", line 99, in convert
    raise ValueError("can not convert %s to %s." % (dtypeobj_in, dtypeobj))
ValueError: can not convert object to float64.

The weird thing is when i tried running it on ubuntu 12.04 desktop using the same installation procedure, it works....

[RawanMG]

I'm getting the same error. Did you by any chance fix it?

[RawanMG]

importing io and using 'matplotlib' fixed the issue

from skimage import io; io.use_plugin('matplotlib')

[ghost]

For me, switching from PIL to pillow fixed a similar error.

[jabkim]

If you use PIL package uninstall it and use pillow.

[ivendrov]

I encountered this error on images in the COCO dataset. Switching from PIL to Pillow alone wasn't enough; I also had to add

from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

as in this StackOverflow answer.