PointCNN Which placeholders for inference ?

Hello,

I would like to test the provided pre-trained model "ScanNet" you put on PoinCNN home page. But I can not find any "inference.py" file. So I would like to implement it on my own, but in order to freeze model, which tensors should I select in the model to do this ?

Thanks!

Oct 10 '19 13:10 jean-noelp

You can select following node as output to freeze pretrained model https://github.com/yangyanli/PointCNN/blob/7d0af994718dd49c66faf21c03c964613e8bdc6f/train_val_cls.py#L177 Thanks

Oct 11 '19 02:10 burui11087

Ok, thanks. And for inputs ? I have just batches of point features (xyz + rgb + nx ny nz) for now.

Oct 11 '19 12:10 jean-noelp

Please refer to following code for inputs https://github.com/yangyanli/PointCNN/blob/fa3b4d46a68450c6e0006b5c0cac014c94398fd2/train_val_cls.py#L172 and modify 'data_dim' to 9 and 'use_extra_features' to True https://github.com/yangyanli/PointCNN/blob/master/pointcnn_cls/scannet_x2_l4.py

Jan 07 '20 09:01 burui11087