Closed apoman38 closed 3 years ago
Getting the following error Havent' u changed the training so that the classifiers extract probabilistic outputs?
Traceback (most recent call last):
File "wrapper.py", line 110, in <module>
main(sys.argv)
File "wrapper.py", line 80, in main
probas, classes = video_class_predict(features, algorithm)
File "wrapper.py", line 58, in video_class_predict
results = model.predict_proba(features_scaled)
File "/Users/tyiannak/.pyenv/versions/3.7.3/lib/python3.7/site-packages/sklearn/utils/metaestimators.py", line 114, in __get__
getattr(delegate, self.attribute_name)
File "/Users/tyiannak/.pyenv/versions/3.7.3/lib/python3.7/site-packages/sklearn/svm/_base.py", line 666, in predict_proba
self._check_proba()
File "/Users/tyiannak/.pyenv/versions/3.7.3/lib/python3.7/site-packages/sklearn/svm/_base.py", line 633, in _check_proba
raise AttributeError("predict_proba is not available when "
AttributeError: predict_proba is not available when probability=False
With which classifier, you trained the model?
With which classifier, you trained the model?
SVM
Please pull and check again. Let me know if there are more issues.
Wrapper.py update: All the outputs are saved in a CSV format file.
How to QA:
Run python3 wrapper.py -i folder -m Algorithm -o output_file.csv
.
Check if the output file is created and works properly.
What this PR does:
1) Add docstrings to functions of shot_generation.py file, usage example, etc. 2) Process video returns more values when shot_generation was created. So added some underscores when the process_video() called. 3) In wrapper.py added posterior probability calculation of every class.
How to QA: 1) Train a model with train.py 2) Run wrapper.py on a single file. Should be printed the posterior probability of every class of the given file. 3) Run wrapper.py in a directory. Should be printed at the end of the process the mean of posteriors. Also, check if a file is created with the name of the directory that you gave. This file contains the posterior probability of every video extensively.