Creating dataset workflow isn't working

[PyxAI]

Hi,
I'm trying to build a dataset using the provided scripts:
get_data.py -> preprocess_audio_frames.py -> face_detector.py -> generate_pickle.py

There are several issues with the pipeline, like (in generate_pickle.py) files are not found in the generated folder (by preprocess_audio_frames \ face_detector)
Yet after overcoming this, I get training issues that result from 0 length wav files, and what seems to be too large files.
In the dataset that you use there are only 32kb file size for the audio.
Any way to give an explanation about what the files should look like, or what causes the errors in training?

[PyxAI]

Here is how to fix it:
round the variable avg_fps on face_detector.py:
avg_fps = np.round(...)
otherwise it will only catch the first frame of the entire video

[upretiii00]

@PyxAI hey! did you face this issue , i have the generated audio files and frames stored and converted them to pickle file using generate_pickle and yet when running runtime.py its saying Total amount of training samples: 0 faces | 0 audios

Here is how to fix it:
round the variable avg_fps on face_detector.py:
avg_fps = np.round(...)
otherwise it will only catch the first frame of the entire video

@PyxAI hey! did you face this issue , i have the generated audio files and frames stored and converted them to pickle file using generate_pickle and yet when running runtime.py its saying Total amount of training samples: 0 faces | 0 audios

[PyxAI]

Did you reference the correct pickle in the settings file?