matanatz/pcnn

Point Convolutional Neural Networks by Extension Operators

pcnn's Issues

Segmentation dataset

Hello,
Thanks for your great work, I want to know where can I download the segmentation data?

Thank you very much!

Segmentation evaluation

Hello,
It seems that the evaluation file for part segmentation is missing?
Also, I find that the segmentation network and classification network are trained independently, I wonder if these two networks can be combined in the convolution stage, then the network can split for different tasks, just like the architecture of pointnet/pointnet++?
Thanks.

How to prepare own data for segmentation

Hi, Since the link http://www.wisdom.weizmann.ac.il/~haggaim/projects/PCNN/segmentation_data.zip is not avaiable now.
How can I prepare my own data.
I can understand
pointclouds_ph = tf.placeholder(tf.float32, shape=(batch_size, point_num, 3))
seg_ph = tf.placeholder(tf.int32, shape=(batch_size, point_num))
but I am confusing about "input_label_ph = tf.placeholder(tf.float32, shape=(batch_size, NUM_CATEGORIES))"
NUM_CATEGORIES seems to be the number of semantic parts.
How to define labels for my own data. Thanks a lot

Enquiry of the normal estimation

Hi @matanatz,

Many thanks for releasing the code.

I was wondering if the normal estimation code is available (even if some files cannot be run)? I am having difficulties in reproducing the normal estimation results in table 4 using PointNet and PointNet++. The best mean cosine loss for pointnet I can get is about 0.65, which is far from 0.47 in the table so I think I must have some dumb mistakes. Table 4 in your PCNN paper is below:

Here are some follow-up questions:

1. I was wondering if you have any advice or any code for normal estimation part I can read?
2. Also, about the cosine loss in the table, is the cosine loss defined as 1 - cos_similarity?
3. About the pointnet part segmentation code, did you use the part segmentation structure in the pointnet main paper (i.e. 1088-channel before FC layers) or did you use the enhanced version of part segmentation structure mentioned in the pointnet supplementary material (i.e. 3024-channel before FC layers)? I have pasted the two structures below:

The original Pointnet part segmentation structure:

The enhanced version part segmentation structure, mentioned in pointnet supplementary:

Best,
Zirui

Cannot run PCNN

Hi,

I think I have all the prerequisites installed:

I'm working on windows.
Can you help?

cp: cannot stat 'segprovider.py': No such file or directory
cp: omitting directory './layers/__pycache__'
--2019-07-13 14:53:47--  https://shapenet.cs.stanford.edu/media/modelnet40_ply_hdf5_2048.zip;
Resolving shapenet.cs.stanford.edu... 171.67.77.19
Connecting to shapenet.cs.stanford.edu|171.67.77.19|:443... connected.
WARNING: cannot verify shapenet.cs.stanford.edu's certificate, issued by `/C=US/ST=MI/L=Ann Arbor/O=Internet2/OU=InCommon/CN=InCommon RSA Server CA':
  Self-signed certificate encountered.
WARNING: certificate common name `cs.stanford.edu' doesn't match requested host name `shapenet.cs.stanford.edu'.
HTTP request sent, awaiting response... 404 Not Found
2019-07-13 14:53:48 ERROR 404: Not Found.

--2019-07-13 14:53:48--  http://unzip/
Resolving unzip... failed: Unknown host.
wget: unable to resolve host address `unzip'
--2019-07-13 14:53:55--  http://modelnet40_ply_hdf5_2048.zip/
Resolving modelnet40_ply_hdf5_2048.zip... 92.242.132.24
Connecting to modelnet40_ply_hdf5_2048.zip|92.242.132.24|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: unspecified [text/html]
Saving to: `index.html.3'

     0K                                                         578K=0.002s

2019-07-13 14:53:55 (578 KB/s) - `index.html.3' saved [918]

FINISHED --2019-07-13 14:53:55--
Downloaded: 1 files, 918 in 0.002s (578 KB/s)
mv: cannot stat 'modelnet40_ply_hdf5_2048': No such file or directory
rm: cannot remove 'modelnet40_ply_hdf5_2048.zip': No such file or directory

tf version : 1.13.1
num point : 1024
experiment name : pcnn
Traceback (most recent call last):
  File "train.py", line 343, in <module>
    train()
  File "train.py", line 42, in train
    TRAIN_FILES = provider.getTrainDataFiles()
  File "D:\CodingProjects\pcnn\provider.py", line 114, in getTrainDataFiles
    return self.getDataFiles(self.train_files)
  File "D:\CodingProjects\pcnn\provider.py", line 117, in getDataFiles
    return [line.rstrip() for line in open(list_filename)]
FileNotFoundError: [Errno 2] No such file or directory: '.\\data\\modelnet40_ply_hdf5_2048\\train_files.txt'
(tensorflow)

Readme file is missing

Hi Matan,

Thanks very much for making your code public. A readme file may be very helpful for users. By the way, whether the provided code can be used for semantic segmentation too?

how to evaluate the accuracy of classificition?

how to evaluate the accuracy of classificition?which folder of evaluate.py can be used to evaluate finial model?