facebookresearch / votenet

Deep Hough Voting for 3D Object Detection in Point Clouds
MIT License
1.69k stars 376 forks source link

Problems with trimesh #40

Open bugdary opened 4 years ago

bugdary commented 4 years ago

Does anyone encounter problems when trying to dump the scene? I installed trimesh = 2.35.39 and I get the following problem :

Finished detection. 37 object detected. Traceback (most recent call last): File "demo.py", line 101, in MODEL.dump_results(end_points, dump_dir, DC, True) File "D:\Projects\Depth\votenet\models\dump_helper.py", line 86, in dump_results pc_util.write_oriented_bbox(obbs[objectness_prob>DUMP_CONF_THRESH,:], os.path.join(dump_dir, '%06d_pred_confident_bbox.ply'%(idx_beg+i))) File "D:\Projects\Depth\votenet\utils\pc_util.py", line 421, in write_oriented_bbox mesh_list = trimesh.util.concatenate(scene.dump()) File "d:\ProgramData\Anaconda3\envs\pytorch_1_1_python36\lib\site-packages\trimesh\scene\scene.py", line 499, in dump for node_name in self.graph.nodes_geometry: File "d:\ProgramData\Anaconda3\envs\pytorch_1_1_python36\lib\site-packages\trimesh\caching.py", line 88, in get_cached value = function(*args, **kwargs) File "d:\ProgramData\Anaconda3\envs\pytorch_1_1_python36\lib\site-packages\trimesh\scene\transforms.py", line 224, in nodes_geometry 'geometry' in self.transforms.node[n]): AttributeError: 'EnforcedForest' object has no attribute 'node'

Any ideas ?

bugdary commented 4 years ago

I solved the issues by upgrading to the latest version of Trimesh and changing the following line:

trimesh.io.export.export_mesh(mesh_list, out_filename, file_type='ply')

to

mesh_list.export(out_filename)

Thanks for sharing your code.

jingnanshi commented 4 years ago

I encountered the same error. After some digging, it looks like it is caused by the fact that the version of networkx installed is too new. For a method that does not involve changing the code: before installing trimesh, do pip install 'networkx>=2.2,<2.3'.

gkd2020 commented 4 years ago

I have a similar problem. But the warning information is different. I don't know what to do. Can someone help me?

Constructed model. Loaded checkpoint /home/votenet-master/demo_files/pretrained_votenet_on_sunrgbd.tar (epoch: 180) Loaded point cloud data: /home/votenet-master/demo_files/input_pc_sunrgbd.ply Inference time: 0.138507 Finished detection. 9 object detected. Traceback (most recent call last): File "demo.py", line 101, in MODEL.dump_results(end_points, dump_dir, DC, True) File "/home/votenet-master/models/dump_helper.py", line 86, in dump_results pc_util.write_oriented_bbox(obbs[objectness_prob>DUMP_CONF_THRESH,:], os.path.join(dump_dir, '%06d_pred_confident_bbox.ply'%(idx_beg+i))) File "/home/votenet-master/utils/pc_util.py", line 421, in write_oriented_bbox mesh_list = trimesh.util.concatenate(scene.dump()) File "/usr/local/lib/python3.6/dist-packages/trimesh/scene/scene.py", line 499, in dump for node_name in self.graph.nodes_geometry: File "/usr/local/lib/python3.6/dist-packages/trimesh/caching.py", line 88, in get_cached value = function(*args, **kwargs) File "/usr/local/lib/python3.6/dist-packages/trimesh/scene/transforms.py", line 224, in nodes_geometry 'geometry' in self.transforms.node[n]): AttributeError: 'EnforcedForest' object has no attribute 'node'

Is there any outstanding scholar who can help me. Thanks

jingnanshi commented 4 years ago

@ai-starter have you tried pip install 'networkx>=2.2,<2.3' before installing trimesh? Your error message seems to be the same.

gkd2020 commented 4 years ago

@ai-starter have you tried pip install 'networkx>=2.2,<2.3' before installing trimesh? Your error message seems to be the same.

Thanks,It is useful! But I have another problem. When I open the results with meshleb, it looks weird. The result is some combination of cubes, which is different from what we see in the paper. what happen? I look forward to your reply. Thanks 2020-02-24 16-47-11屏幕截图

jingnanshi commented 4 years ago

You need to change meshlab to only show the wireframes. Also, please open a new issue for this.