Training with all the frames & anomaly detection

[santhoshkelathodi]

Assuming this is the implementation of the paper "Abnormal Event Detection in Videos using SpatioTemporal AutoEncoder"

1. I think all frames need to be used for training than sampled frames. The parameter fps does not reflect the correct operation. What if I want to sample with the parameter value less than 1
2. As I understand even for testing the processor.py needs to be run to create the test.npy file. The process may be mentioned in the readme

[santhoshkelathodi]

I have created a code for extracting all the frames from the train and test. Can it be uploaded here?

[santhoshkelathodi]

I have 2 questions: 1)I could not understand why the step: #Reshape to (227,227,batch_size) imagestore.resize(b,c,a) is required 2) in the training batch size is fixed as 1. What is the influence of changing it to higher values?

[BelivLi]

can you upload the code for extracting all the frames? thanks And i have one question how to generate the test.npy file,Is not changing the training section in the processor.py file to testing?

[ronaldosaravana007]

PLEASE SEND CODE FOR EXTRACTING ALL FRAMES

[ronaldosaravana007]

I have created a code for extracting all the frames from the train and test. Can it be uploaded here?

BRO send me that code in my mail [email protected]