ValueError: file_type not specified!

[ut-austin-rpl]

ValueError: file_type not specified!

[Steve-Tod]

It seems the problem is with trimesh. First, try upgrading trimesh. If it is still not working, can you print out the self.mesh_dir / (name + "_aff.obj") and debug? Or just specify the file_type=obj'.

[ccc1711]

I update trimesh to the latest version and it still doesn’t work, and I print out the path of self.mesh_dir / (name + "_aff.obj"),is PosixPath(‘data/experiments/21-08-13-20-33-46/meshes/round_000_trial_000_aff.obj') .How to specify the file_type='obj' ?

[Steve-Tod]

You can try aff_mesh.export(self.mesh_dir / (name + "_aff.obj"), file_type='obj'). If this still does not work, actually the visualization is not necessary for running, and you can remove the argument --vis.

[wuzeww]

I update trimesh to the latest version and it still doesn’t work, and I print out the path of self.mesh_dir / (name + "_aff.obj"),is PosixPath(‘data/experiments/21-08-13-20-33-46/meshes/round_000_trial_000_aff.obj') .How to specify the file_type='obj' ?

Hi，
excuse me, have you realized the function of visualization?

[ccc1711]

I update trimesh to the latest version and it still doesn’t work, and I print out the path of self.mesh_dir / (name + "_aff.obj"),is PosixPath(‘data/experiments/21-08-13-20-33-46/meshes/round_000_trial_000_aff.obj') .How to specify the file_type='obj' ?

Hi，
excuse me, have you realized the function of visualization?

Not yet, how about you?

[wuzeww]

You can try aff_mesh.export(self.mesh_dir / (name + "_aff.obj"), file_type='obj'). If this still does not work, actually the visualization is not necessary for running, and you can remove the argument --vis.

Excuse me, I want to achieve the visualization results (Visualization of the grasp affordance landscape and predicted grasps) in your paper. Does the code you upload have this function?

[wuzeww]

I update trimesh to the latest version and it still doesn’t work, and I print out the path of self.mesh_dir / (name + "_aff.obj"),is PosixPath(‘data/experiments/21-08-13-20-33-46/meshes/round_000_trial_000_aff.obj') .How to specify the file_type='obj' ?

Hi，
excuse me, have you realized the function of visualization?

Not yet, how about you?

not yet

[Steve-Tod]

You can try aff_mesh.export(self.mesh_dir / (name + "_aff.obj"), file_type='obj'). If this still does not work, actually the visualization is not necessary for running, and you can remove the argument --vis.

Excuse me, I want to achieve the visualization results (Visualization of the grasp affordance landscape and predicted grasps) in your paper. Does the code you upload have this function?

Yes, the code will save the visualization to obj file if you use the argument --vis. However, I don't know why you meet error when running aff_mesh.export.

[wuzeww]

I tried aff_mesh.export(str(self.mesh_dir / (name + "_aff.obj"))), the issue of file_type disappeared.

Then i appeared the issue of array out of bounds (IndexError: index 201 is out of bounds for axis 0 with size 201).

Finally, found that the data of data/urdfs/packed/test/Paprika_800_tex_visual.obj is short of the letter 'v'(at the line of 15).

[Steve-Tod]

I tried aff_mesh.export(str(self.mesh_dir / (name + "_aff.obj"))), the issue of file_type disappeared.

Then i appeared the issue of array out of bounds (IndexError: index 201 is out of bounds for axis 0 with size 201).

Finally, found that the data of data/urdfs/packed/test/Paprika_800_tex_visual.obj is short of the letter 'v'(at the line of 15).

Hmm, that's interesting. Because I checked the data that I uploaded, and there's no such problem.

[wuzeww]

I tried aff_mesh.export(str(self.mesh_dir / (name + "_aff.obj"))), the issue of file_type disappeared.
Then i appeared the issue of array out of bounds (IndexError: index 201 is out of bounds for axis 0 with size 201).
Finally, found that the data of data/urdfs/packed/test/Paprika_800_tex_visual.obj is short of the letter 'v'(at the line of 15).

Hmm, that's interesting. Because I checked the data that I uploaded, and there's no such problem.

I just re-extracted the data.zip file and found that the data is complete.

I want to know how long have you cost about training vgn one epoch. I performed on a computer with a GTX 1080 Ti graphics card and the memory is 32G, but my cpu is noot good. I cost about 14 hours one epoch. I want to know if it’s the cpu problem.

[Steve-Tod]

I tried aff_mesh.export(str(self.mesh_dir / (name + "_aff.obj"))), the issue of file_type disappeared.
Then i appeared the issue of array out of bounds (IndexError: index 201 is out of bounds for axis 0 with size 201).
Finally, found that the data of data/urdfs/packed/test/Paprika_800_tex_visual.obj is short of the letter 'v'(at the line of 15).

Hmm, that's interesting. Because I checked the data that I uploaded, and there's no such problem.

I just re-extracted the data.zip file and found that the data is complete.

I want to know how long have you cost about training vgn one epoch. I performed on a computer with a GTX 1080 Ti graphics card and the memory is 32G, but my cpu is noot good. I cost about 14 hours one epoch. I want to know if it’s the cpu problem.

When training the network on pile dataset (the larger one), one epoch takes 2 hours on one RTX 2080 Ti. So I think the problem is with CPU. You can check the "Volatile GPU-Util" given by nvidia-smi to see if the GPU is fully used. GPU-Util over 80% is a good sign.