Install Pytorch tensorflow 1.3.0 tensorflow-tensorboard 0.1.5 torchvision 0.1.8 tensorboard 2.2.2 pytorch 1.7.1 mayavi 4.7.3.dev0 /4.5.0 traits 6.1.1 traitsui 5.1.0 vtk 8.1.2 p pyqt 4.11.4 pyqt5 5.15.2 qt 4.8.5
The code was trained by PCPNet Dataset
To train a model run "train_multi_experts3_atten.py". You also can use "stacked_with_dgcnn_experts3_normal_estimation_nopca2_with_atten256/512/1024" to test for a patch with 256/512/1024 points.
you can obtain the estimation results via BaiduYun link: https://pan.baidu.com/s/1ZDx8TwS9AmJTXxoS1PkePQ code: 673a The code is built on the top of PCPNet and DGCNN. Thanks for the precedent contribution.