How to obtain point cloud files in point cloud completion testing #21

yizhiboyan · 2023-04-25T15:50:53Z

Thank you for your excellent work! I am a beginner who trained the point cloud completion task using your code and obtained the checkpoint and log files in the exp folder. How can I obtain the completed point cloud file (. pcd) generated during testing? Looking forward to your reply.

AllenXiangX · 2023-04-27T01:48:15Z

Hi,
You can use the Imdb reader of GRNet to get the .pcd files from the PCN dataset.

yizhiboyan · 2023-04-27T06:41:50Z

Thank you for your patient reply. I will give it a try,

huyanbi · 2023-05-03T07:31:14Z

@yizhiboyan ,Hello, I understand from your answer that you have successfully reproduced the point cloud completion. Have you encountered this problem?Loaded compiled 3D CUDA chamfer distance 0%| | 0/906 [00:03<?, ?it/s] Traceback (most recent call last): File "train.py", line 167, in <module> train(config) File "train.py", line 105, in train pcds_pred = model(partial) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/model_completion.py", line 137, in forward feat = self.feat_extractor(point_cloud) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/model_completion.py", line 32, in forward l1_xyz, l1_points, idx1 = self.sa_module_1(l0_xyz, l0_points) # (B, 3, 512), (B, 128, 512) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/utils.py", line 375, in forward new_xyz, new_points, idx, grouped_xyz = sample_and_group_knn(xyz, points, self.npoint, self.nsample, self.use_xyz, idx=idx) File "../models/utils.py", line 316, in sample_and_group_knn new_xyz = gather_operation(xyz, furthest_point_sample(xyz_flipped, npoint)) # (B, 3, npoint) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/pointnet2_ops-3.0.0-py3.7-linux-x86_64.egg/pointnet2_ops/pointnet2_utils.py", line 54, in forward out = _ext.furthest_point_sampling(xyz, npoint) RuntimeError: false INTERNAL ASSERT FAILED at "pointnet2_ops/_ext-src/src/sampling.cpp":83, please report a bug to PyTorch. CPU not supported
Looking forward to your answer

yizhiboyan · 2023-05-04T03:22:06Z

@yizhiboyan ,Hello, I understand from your answer that you have successfully reproduced the point cloud completion. Have you encountered this problem?Loaded compiled 3D CUDA chamfer distance 0%| | 0/906 [00:03<?, ?it/s] Traceback (most recent call last): File "train.py", line 167, in <module> train(config) File "train.py", line 105, in train pcds_pred = model(partial) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/model_completion.py", line 137, in forward feat = self.feat_extractor(point_cloud) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/model_completion.py", line 32, in forward l1_xyz, l1_points, idx1 = self.sa_module_1(l0_xyz, l0_points) # (B, 3, 512), (B, 128, 512) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "../models/utils.py", line 375, in forward new_xyz, new_points, idx, grouped_xyz = sample_and_group_knn(xyz, points, self.npoint, self.nsample, self.use_xyz, idx=idx) File "../models/utils.py", line 316, in sample_and_group_knn new_xyz = gather_operation(xyz, furthest_point_sample(xyz_flipped, npoint)) # (B, 3, npoint) File "/usr/local/miniconda3/envs/spd/lib/python3.7/site-packages/pointnet2_ops-3.0.0-py3.7-linux-x86_64.egg/pointnet2_ops/pointnet2_utils.py", line 54, in forward out = _ext.furthest_point_sampling(xyz, npoint) RuntimeError: false INI haven't encountered this problem beforeTERNAL ASSERT FAILED at "pointnet2_ops/_ext-src/src/sampling.cpp":83, please report a bug to PyTorch. CPU not supported Looking forward to your answer

Hi,I haven't encountered this problem before,because I use https://github.com/AllenXiangX/SPD_jittorSPD_jittor.This version uses jittor, similar to pytorch,use python,you could try this.This is my environment configuration information, I hope it can be helpful to you.I use ubuntu20.04LTS

packages in environment at /home/usrname/anaconda3/envs/spd_jittor:

yizhiboyan · 2023-05-04T06:30:12Z

Hi, You can use the Imdb reader of GRNet to get the .pcd files from the PCN dataset.

Hi,I have encountered a new problem.

When I train the completion task,use SPD_jittor, I get the optimizer: none,
and I debug in train, py, about 128 lines

print('epoch: ', epoch_idx, 'optimizer: ', optimizer.param_groups[0].get('lr'))

Is there any problem with the parameter lr not present in param_groups [0]?
Looking forward to your reply.

by0619 · 2024-01-19T02:48:52Z

Thank you for your excellent work! I am a beginner who trained the point cloud completion task using your code and obtained the checkpoint and log files in the exp folder. How can I obtain the completed point cloud file (. pcd) generated during testing? Looking forward to your reply.

@yizhiboyan Hello, I have also obtained the checkpoint and log files and I tried to save the test output point cloud as author advised, but I have difficulty in storing and obtaining the output completed point cloud(.pcd) generated through testing , would you please share your code for obtaining output point cloud? Thanks a lot and great appreciation!

yizhiboyan · 2024-01-19T08:32:30Z

Hello,you would like to try this.
test.py
#if you have not outputfile folders
if not config.test.out_path:
config.test.out_path = "./outpcdfiles"
output_dir = os.path.join(config.test.out_path, datetime.now().isoformat())

with tqdm(test_dataloader) as t:
    for model_idx, (taxonomy_id, model_id, data) in enumerate(t):
        taxonomy_id = taxonomy_id[0] if isinstance(taxonomy_id[0], str) else taxonomy_id[0].item()
        model_id = model_id[0]


        if config.dataset.name  in ['PCN', 'Completion3D']:
            with torch.no_grad():
                for k, v in data.items():
                    data[k] = helpers.var_or_cuda(v)

                partial = data['partial_cloud']
                gt = data['gtcloud']

                b, n, _ = partial.shape

                pcds_pred = model(partial.contiguous())           

                pcds_out = pcds_pred[-1]                       #pcds_out is 3rd stage completion cloud 
                pcds_out_num = pcds_out.cpu().numpy()           
                pcds_out_num = pcds_out_num.reshape(-1, 3)     #trans to numpy
                complete_output_folder = os.path.join(output_dir, taxonomy_id)
                os.makedirs(complete_output_folder, exist_ok=True)

                if config.dataset.id == "pcn":
                    pcd = o3d.geometry.PointCloud()
                    pcd.points = o3d.utility.Vector3dVector(pcds_out_num)

                    o3d.io.write_point_cloud(os.path.join(complete_output_folder, '%s.ply' % model_id), pcd)


                    loss_total, losses = completion_loss.get_loss(pcds_pred, partial, gt)

                    partial_matching = losses[0].item() * multiplier
                    loss_c = losses[1].item() * multiplier
                    loss_1 = losses[2].item() * multiplier
                    loss_2 = losses[3].item() * multiplier
                    loss_3 = losses[4].item() * multiplier
                    ……………………

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

How to obtain point cloud files in point cloud completion testing #21

How to obtain point cloud files in point cloud completion testing #21

yizhiboyan commented Apr 25, 2023

AllenXiangX commented Apr 27, 2023

yizhiboyan commented Apr 27, 2023

huyanbi commented May 3, 2023

yizhiboyan commented May 4, 2023

yizhiboyan commented May 4, 2023

by0619 commented Jan 19, 2024

yizhiboyan commented Jan 19, 2024

How to obtain point cloud files in point cloud completion testing #21

How to obtain point cloud files in point cloud completion testing #21

Comments

yizhiboyan commented Apr 25, 2023

AllenXiangX commented Apr 27, 2023

yizhiboyan commented Apr 27, 2023

huyanbi commented May 3, 2023

yizhiboyan commented May 4, 2023

packages in environment at /home/usrname/anaconda3/envs/spd_jittor:

Name Version Build Channel

yizhiboyan commented May 4, 2023

by0619 commented Jan 19, 2024

yizhiboyan commented Jan 19, 2024