I think you are not doing filtration on ECG images because you already used filtered database (ECG-ID). So, if I am generating my own filtered ECG database (ECG images for each person) to be trained here without errors, what are the constraints (# of samples, for example) on the uploaded images?
Hello,
I'm experiencing the following issues when trying to use train.py:
How should I modify data_generator.py to resolve this issue?
Additionally, it appears that I can't use model_ecg.h5, so I would like to try retraining.
The "Shuffle()" in make_one_shot_task function returns an error. Are we supposed to import it some package for it or is it a implemented function? I tried shuffling the dataset with numpy and the accuracy is 18% at 3000 iterations wheras it is mentioned in the blog that the accuracy reached is 85%
shuffle is not defined ,if i use "from import import shuffle" ,still error βshuffle() takes from 2 to 3 positional arguments but 4 were givenβγ
so i would like to ask how you used the "shuffle"