Deeplabcut: Unable to Label Videos: ImportError: Missing optional dependency 'tables'. Use pip or conda to install tables.

[linussun]

* Bug report **
Description: Unable to label videos due to error generated:
"ImportError: Missing optional dependency 'tables'. Use pip or conda to install tables."

To Reproduce
Windows 10
Python=3.6
tensorflow-gpu==1.13.1
pytables==3.4.3
numpy==1.17.2

Expected behavior
I expected a labelled video to be written.

* Error Output *

`---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-8-560a625bda99> in <module>
      6 deeplabcut.convertcsv2h5(path_config_file)
      7 print('Starting Create Labelled Video')
----> 8 deeplabcut.create_labeled_video(path_config_file,videofile_path,videotype='mp4')
      9 #deeplabcut.create_labeled_video(path_config_file,videofile_path,videotype='mp4')

C:\ProgramData\Anaconda3\envs\dlc-GPU-C10t13\lib\site-packages\deeplabcut\utils\make_labeled_video.py in create_labeled_video(config, videos, videotype, shuffle, trainingsetindex, filtered, save_frames, Frames2plot, delete, displayedbodyparts, codec, outputframerate, destfolder, draw_skeleton, trailpoints, displaycropped)
    376         else:
    377             print("Loading ", video, "and data.")
--> 378             datafound,metadata,Dataframe,DLCscorer,suffix=auxiliaryfunctions.LoadAnalyzedData(str(videofolder),vname,DLCscorer,filtered) #returns boolean variable if data was found and metadata + pandas array
    379             videooutname=os.path.join(vname + DLCscorer+suffix+'_labeled.mp4')
    380             if datafound and not os.path.isfile(videooutname): #checking again, for this loader video could exist

C:\ProgramData\Anaconda3\envs\dlc-GPU-C10t13\lib\site-packages\deeplabcut\utils\auxiliaryfunctions.py in LoadAnalyzedData(videofolder, vname, DLCscorer, filtered)
    447                 DLCscorer='DLC_'+(datanames[0].split('DLC_')[1]).split('.h5')[0]
    448             print("Creating output for:", DLCscorer," instead.")
--> 449             Dataframe = pd.read_hdf(datanames[0])
    450             metadata=LoadVideoMetadata(datanames[0])
    451             datafound=True

C:\ProgramData\Anaconda3\envs\dlc-GPU-C10t13\lib\site-packages\pandas\io\pytables.py in read_hdf(path_or_buf, key, mode, errors, where, start, stop, columns, iterator, chunksize, **kwargs)
    395             raise FileNotFoundError(f"File {path_or_buf} does not exist")
    396 
--> 397         store = HDFStore(path_or_buf, mode=mode, errors=errors, **kwargs)
    398         # can't auto open/close if we are using an iterator
    399         # so delegate to the iterator

C:\ProgramData\Anaconda3\envs\dlc-GPU-C10t13\lib\site-packages\pandas\io\pytables.py in __init__(self, path, mode, complevel, complib, fletcher32, **kwargs)
    516             raise ValueError("format is not a defined argument for HDFStore")
    517 
--> 518         tables = import_optional_dependency("tables")
    519 
    520         if complib is not None and complib not in tables.filters.all_complibs:

C:\ProgramData\Anaconda3\envs\dlc-GPU-C10t13\lib\site-packages\pandas\compat\_optional.py in import_optional_dependency(name, extra, raise_on_missing, on_version)
     90     except ImportError:
     91         if raise_on_missing:
---> 92             raise ImportError(msg) from None
     93         else:
     94             return None

ImportError: Missing optional dependency 'tables'.  Use pip or conda to install tables.`

Desktop (please complete the following information about your system):

• OS: Windows10
• DeepLabCut Version: 2.1.5.2
• Browser Chrome

Additional context
I tried trying older solutions including pip uninstall numpy, and conda install numpy==1.17.2 and I'm still getting this tables error.

[MMathisLab]

Seems your tables is broken. There are several issues on this and the Troubleshooting Tips: https://github.com/AlexEMG/DeepLabCut/wiki/Troubleshooting-Tips If you use the search fucntion in this repo yuo will find several solutions.

Windows: one user reports an issue with PyTables (i.e. here) pd.read_hdf(path). The solution was to upgrade tables from 3.4.4 to 3.4.5.dev using: pip install git+https://github.com/PyTables/PyTables

https://github.com/AlexEMG/DeepLabCut/issues/431

https://github.com/AlexEMG/DeepLabCut/issues/377

https://github.com/AlexEMG/DeepLabCut/issues/63

[linussun]

Thank you for the advice!

maceng4's solution worked for me to get past the pytables error (#63 ) https://github.com/AlexEMG/DeepLabCut/issues/63

Still can't label videos at the moment, now on to float division by zero, ffmpeg opencv error!

Other information: I used the default .yaml Windows GPU config file to install on a vanilla anaconda install, Windows 10 (1903).

[MMathisLab]

maybe check out this. We also just added code to check video for errors, so that will be in 2.1.6; and/or you can directly import and use the source code and this will be a more useful warning... see:

https://github.com/AlexEMG/DeepLabCut/issues/557

Yes, sadly windows can have so many ways to install things/depends on what was installed already, as conda isn't a real virtual env; ie like docker, so there can be a lot of little differences in how/where things get installed (shout out for ubuntu if you want to take the plunge!)

[linussun]

Maybe I'll go back to Ubuntu, I'm not afraid of it just dependent on Windows Apps :(

Also, I figured out the video issue, it was the pathname after all, it must be explicit rather than relative.

Thanks for the quick replies!