Training 2 actions

[GyungBin]

Hello! Thank you for sharing your work! There was an error when running s5_test.py via webcam after training its own dataset with two actions.

File "C:\Realtime-Action-Recognition-master\utils\lib_classifier.py", line 185, in smooth_scores
    score_sums += score
ValueError: operands could not be broadcast together with shapes (9,) (2,) (9,) 

I think it's related to the matrix, could you help me that?

[Amyhds]

Hi! I have the same problem. how did you solve it?

[qkdkralsgh]

@Amyhds Hi! maybe you need to set window_size & classses

[vpeopleonatank]

I also faced this issue because I use the original model which is trained for 9 classes, but my new dataset has 6 classes. When run training script at step 4, my model was not saved because the qt error occured before saving model. You should check classes in config.yaml file or trained model file.

[Amyhds]

#classes: ['stand', 'walk', 'run', 'jump', 'sit', 'squat', 'kick', 'punch', 'wave']
classes: ['lying', 'other', 'sitting', 'standing']

I changed classes like that in config.yaml but why should I change window_size? Could you tell me why? what else do I have to change?

[qkdkralsgh]

@Amyhds If you changed window_size, you need to set window_size. When training the model, the pca, window_size, and classes must be set well.

[Amyhds]

Thanks for the answer!!!