AGOberprieler / GinJinn2

https://ginjinn2.readthedocs.io
Apache License 2.0
15 stars 2 forks source link

pickle.UnpicklingError in ginjinn predict #3

Closed mareikedaubert closed 1 year ago

mareikedaubert commented 2 years ago

Hello, I have been following the example for detecting leaf outlines from scanned herbarium sheets and applied it to my own images. However, I run into an error when trying to predict from a segmentation model:

  File "/user/me/.conda/envs/ginjinn2/lib/python3.7/site-packages/torch/serialization.py", line 595, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/user/me/.conda/envs/ginjinn2/lib/python3.7/site-packages/torch/serialization.py", line 764, in _legacy_load
    magic_number = pickle_module.load(f, **pickle_load_args)
_pickle.UnpicklingError: invalid load key, '<'.

The error does not occur for predicting from a bounding box model and not when I run ginjinn predict without the -r option. (Apart from that the application seems to work fine for my herbarium sheets with the workflow described in the example). I run the pipeline on my university's HPC in a conda environment with python 3.7 and cudatoolkit 10.1.

ulilautenschlager commented 1 year ago

The problem has been fixed with v1.0.7.