search

ranahanocka / MeshCNN

Convolutional Neural Network for 3D meshes in PyTorch
MIT License
1.56k stars 314 forks source link

Model too large #109

Open pvnieo opened 3 years ago

pvnieo commented 3 years ago

Hi,

Thanks for the great contribution and for sharing the code.

I'm using MeshCNN as a feature extractor, so I'm using an architecture similar to the one used in segmentation, and because I want features for each edges, I can't simplify the meshes to have fewer edges.

I have meshes with 15k edges, and using these meshes results in a very large model (the model actually have fewer parameters, but the operations take a lot of memory), and very slow iterations.

Is there some suggestions to accelerate the training of the model and also reduce it's size.

Thank you in advance!

AmorDiana commented 5 months ago

Did u find the answer? I had same question.. @pvnieo

pvnieo commented 5 months ago

@AmorDiana no...