search

paschalidoud / superquadric_parsing

Code for "Superquadrics Revisited: Learning 3D Shape Parsing beyond Cuboids", CVPR 2019
https://superquadrics.com
Other
135 stars 34 forks source link

Inquire about training the probability existence #10

Closed mahmoudEltaher closed 3 years ago

mahmoudEltaher commented 3 years ago

Kindly I want to inquire about the probability existence

In the paper ,it said that the probability is trained after 20000 epochs ,how i can find this condition in the code ?

paschalidoud commented 3 years ago

Hi,

You need to specify the number of epochs. Please have a look at the arguments provided in the train_network.py script.

Best, Despoina

mahmoudEltaher commented 3 years ago

Hi @paschalidoud

Thanks for your collaboration.

I know how i can specify the number of epochs.

But i'm asking about this part in the paper "To further increase parsimony, we then fix all parameters except γ for additional 5k iterations. This step removes remaining overlapping primitives as also observed in [37]."

how i can find this condition in the code ?