search

RomeroBarata / skeleton_based_anomaly_detection

Code for the CVPR'19 paper "Learning Regularity in Skeleton Trajectories for Anomaly Detection in Videos"
132 stars 49 forks source link

Different layers while loading model #26

Open Mishalfatima opened 3 years ago

Mishalfatima commented 3 years ago

Hi! I followed your procedure to train and evaluate the model but with my own dataset., The problem is at evaluation time, I get an error like this: "ValueError: You are trying to load a weight file containing 20 layers into a model with 12 layers." Can you please tell me what I might be doing wrong while training or evaluating.

RomeroBarata commented 3 years ago

Hi @Mishalfatima,

Sorry for the delay in my response. Please have a look at this resolved issue https://github.com/RomeroBarata/skeleton_based_anomaly_detection/issues/6.

Kind regards, Romero

bing-0906 commented 3 years ago

Hi@RomeroBarata , thanks for your excellent work. During reimplementation, I also get this error. Based on your suggestion, the training phase and evaluation phase are combined_model.But it still not works. Is there any other potential error?

RomeroBarata commented 3 years ago

Hi @bing-0906,

That's correct, training and evaluation should match model-wise. Sorry but I cannot recall any other source of error. You can try to debug the training code just before the model saving function to make sure that what is being saved is what you would expect during evaluation time.

kuldeepbiradar1 commented 3 years ago

Hi, I have used combined model for evaluation but still same error.

aytek03 commented 2 years ago

Hi, @kuldeepbiradar1

Could you solve this problem?

zdl12345678 commented 1 year ago

I also encountered this problem, has anyone solved it? @RomeroBarata @Mishalfatima @bing-0906 @aytek03 @kuldeepbiradar1