PointASNL
PointASNL copied to clipboard
yanx27
PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling （CVPR 2020）
This aspect is not discussed in the paper. Can anyone share the number of parameter, FLOPs and/or other measures of model complexity of PointASNL for classification and segmentation? Thanks.
I meeting the follow error,can you help me ? omnisky@omnisky:~/Documents/PointASNL-master/tf_ops$ **sh compile_ops.sh** omnisky@omnisky:~/Documents/PointASNL-master/tf_ops$ cd ../ omnisky@omnisky:~/Documents/PointASNL-master$ **python train.py --data /home/omnisky/Documents/PointASNL-master/data/modelnet40_ply_hdf5_2048 --exp_dir PointASNL_without_noise** Traceback (most recent call last): File "train.py", line...
Why can't train_l1_points and train_l2_points tensors be converted to numpy arrays in this code？I would be grateful if you could take time to answer me！！！ `` train_l1_xyz, train_l1_points = PointASNLSetAbstraction(train_l0_xyz,...
The code shows that the PointASNL classification network adopts the first 1024 points as input by default, I am really confused about why the first 1024 points of the aligned...
in AS module, i found that the update of the point is [x, y,z] = k[x, y, z], why?
compile_op.sh have been successfully compiled ，this problem appear when run train.py `InvalidArgumentError (see above for traceback): No OpKernel was registered to support Op 'FarthestPointSample' used by node layer1/FarthestPointSample (defined at...
Hy ! your work is very interesting. I get the following error while trying to run train.py for classification ImportError: No module named nearest_neighbors.lib.python.nearest_neighbors  can you please highlight the...
在AS模块中，您曾说到为了减少计算量，关系函数R为两点的点积相似度，fi转置与fj进行点乘，再除根号D，进行softmax操作，结合代码我不是很了解，这部分代码中，transformed_feature1和trainsformed_new_point都是通过new_point卷积得来的，我理解的这两部分都是近邻点的特征得来的，而没有使用到采样点的特征。能否指点我一二，非常感谢~
