search

ranahanocka / MeshCNN

Convolutional Neural Network for 3D meshes in PyTorch
MIT License
1.58k stars 316 forks source link

how to get mean & std for edge features? #20

Closed TerenceCYJ closed 5 years ago

TerenceCYJ commented 5 years ago

Hi @ranahanocka In__getitem__() meta['edge_features'] = (edge_features - self.mean) / self.std Which self.mean and self.std are read from 'mean_std_cache.p' file.

How to get such file? How to implement it in my own data without such file?

Thanks

ranahanocka commented 5 years ago

If it doesn't exist, it is calculated automatically from the training data. Please organize your data in the same way that I organized it in the example sets, and delete the mean_std_cache.p, and re-run the code.

TerenceCYJ commented 5 years ago

Hi @ranahanocka , Why do you use mean and std here? If it is used for normalization, why do you use the mean and std of features from all samples?

Thanks

ranahanocka commented 5 years ago

@TerenceCYJ It is common practice in machine learning to do something called "input feature normalization", where the distribution of all features in the set is normalized to have mean zero and standard deviation 1. For an explanation, see here.

TerenceCYJ commented 5 years ago

Sure, thank you very much! @ranahanocka