Error in running label-maker for skynet-train

[emackres]

Great to see this new data prep workflow for Skynet.
https://github.com./developmentseed/label-maker/blob/master/examples/skynet-train-data-prep.md

After using label-maker, I got the utility script to run but got an error:

Traceback (most recent call last):
  File "skynet.py", line 24, in <module>
    img = Image.fromarray(label.astype(np.uint8))
  File "/Users/eric/anaconda3/lib/python3.6/site-packages/PIL/Image.py", line 2431, in fromarray
    size = shape[1], shape[0]
IndexError: tuple index out of range

It would be great to know if this is a problem with my setup, or if others are experiencing it and to get it addressed.

Also, I think there is a typo and the reference to the utility script python utils/segnet.py should actually be python utils/skynet.py

[drewbo]

@emackres thanks for trying it out! Can you post your config.json and I'll try to recreate?

[yulifengwx]

@drewbo @emackres I also got the same error. my config.json like this:
{ "country": "togo", "bounding_box": [1.09725, 6.05520, 1.34582, 6.30915], "zoom": 12, "classes": [ { "name": "Buildings", "filter": ["has", "building"] }, { "name": "Road", "filter": ["has", "highway"] } ], "imagery": "http://a.tiles.mapbox.com/v4/mapbox.satellite/{z}/{x}/{y}.jpg?access_token=access-tocken", "background_ratio": 1, "ml_type": "classification" }
the error like this:

Traceback (most recent call last):
File "E:/work/AI/label-maker/examples/utils/skynet.py", line 25, in
img = Image.fromarray(label.astype(np.uint8))
File "D:\ProgramData\Anaconda3\envs\skynet\lib\site-packages\PIL\Image.py", line 2431, in fromarray
size = shape[1], shape[0]
IndexError: tuple index out of range

the cause is that the length of label.shape is 1 , in PIL method Image.fromarray , "size = shape[1], shape[0]" shape[1] index 1 out of range

[drewbo]

@yulifengwx this script is only intended for segmentations labels. Can you try again with the ml_type changed to segmentation?

[yulifengwx]

@yulifengwx this script is only intended for segmentations labels. Can you try again with the ml_type changed to segmentation?
@drewbo Thanks for reply , set ml_type to segmentation solve the error. But this is another error when runing with lastest verison numpy:
skynet.py in line 34, in keys.remove(key) ==> 'KeysView' object has no attribute 'remove'.
with numpy v1.13.3 the error did not occur.

[drewbo]

we're pinned against v1.13.3 for now but let me know if you can make this work for more recent versions