Issue with manual frame extraction

I have a 8-bit multi-tif stack (49 slices) in my video folder for which I want to annotate every frame.
Downscale.tif (2.4 MB)

I tried running the demo-notebook for my own data but I cant go further than the frame extraction.

The call to deeplabcut.extract_frames(path_config_file,'manual', crop=False)
returns the error

Traceback (most recent call last):

  File "C:\Anaconda3\envs\dlc-windowsGPU\lib\site-packages\deeplabcut\generate_training_dataset\", line 355, in browseDir

OverflowError: argument 1 overflowed: value must be in the range -2147483648 to 2147483647

Also the frame annotation tool does not seem to have loaded the stack properly: the slider ranges from 0 to 1 eventhough I have 49 slices in the stack.

extract frames does not support tiff (althogh openCV might). If you want to annotate all frames, you could just extract them as pngs yourself and place them as labeled-data/videoname/img00.png, etc. [also add the videoname to the videolist in the config.yaml file]

Hi thanks for the quick reply !

Ok too bad, is it documented somewhere ? Tiff is quite popular scientific format for microscopy so I would expect similar claims in the future…
Maybe opening an issue, even if it is not meant to be fix at least people can spot it quickly, or just a raise statement in the function when tif are used.

Yes I could add a piece of python code in my notebook for that I guess :sweat_smile:

Sure, tiff/or even folders of images could be added. However, in some sense it’s for extracting frames from videos. Of course this could be added in the future. Feel free to open an issue, and let’s see if somebody contributes! Perhaps you are interested :wink: (and once you write the code for your project that should be quite easy)