codes running issues

[Z-Lee-corder]

Hi, I tried to tun your source code，but the following error occurred：“AttributeError: 'Tensor' object has no attribute 'isnan”
The wrong sentence is “points_out_of_range = ((xyz_local_grid < 0) | (xyz_local_grid >= grid_size) | (xyz_local_grid.isnan())) .any(-1).flatten()”in the folder “pcdet/utils/density_utils.py".

What is the cause and how to improve it?

[jskhu]

Could you provide me with pip freeze and the whole stack trace?

[Z-Lee-corder]

I use the following version: python=3.6 pytorch=1.5.0 and cuda 10.1
Are the versions of these packages too low to cause this error?

[Z-Lee-corder]

I have used the same environment as you mentioned, but the error still occurred, the whole stack trace is following:
Traceback (most recent call last):
File "/media/lizheng/datum/codes/PDV/tools/train_utils/train_utils.py", line 93, in train_model
dataloader_iter=dataloader_iter
File "/media/lizheng/datum/codes/PDV/tools/train_utils/train_utils.py", line 38, in train_one_epoch
loss, tb_dict, disp_dict = model_func(model, batch)
File "/media/lizheng/datum/codes/PDV/pcdet/models/init.py", line 42, in model_func
ret_dict, tb_dict, disp_dict = model(batch_dict)
File "/home/lizheng/anaconda3/envs/pcdet/lib/python3.6/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "/media/lizheng/datum/codes/PDV/pcdet/models/detectors/pdv.py", line 11, in forward
batch_dict = cur_module(batch_dict)
File "/home/lizheng/anaconda3/envs/pcdet/lib/python3.6/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "/media/lizheng/datum/codes/PDV/pcdet/models/roi_heads/pdv_head.py", line 252, in forward
positional_input = self.get_positional_input(batch_dict['points'], batch_dict['rois'], local_roi_grid_points)
File "/media/lizheng/datum/codes/PDV/pcdet/models/roi_heads/pdv_head.py", line 213, in get_positional_input
return_centroid=self.pool_cfg.ATTENTION.POSITIONAL_ENCODER == 'density_centroid')
File "/media/lizheng/datum/codes/PDV/pcdet/utils/density_utils.py", line 80, in find_num_points_per_part_multi
points_out_of_range = ((xyz_local_grid < 0) | (xyz_local_grid >= grid_size) | (xyz_local_grid.isnan())).any(-1).flatten()
AttributeError: 'Tensor' object has no attribute 'isnan'
python-BaseException

[jskhu]

Have you tried running the provided tests in PDV/tests, specifically this test suite?

https://github.com/TRAILab/PDV/blob/main/tests/utils/test_density_utils.py#L120-L279

And to confirm, you have the same environment as follows?

• Linux (tested on Ubuntu 18.04)
• Python 3.8
• PyTorch 1.7
• CUDA 10.2

If it still doesn't work, I would suggest trying to run the code with the provided docker image.

[jskhu]

I also saw another Github issue that may help you solve your problem (yuantn/MI-AOD#2 (comment)). The code change seemed to work fine on my end, but I haven't tested this enough to guarantee that it will also work.

[Z-Lee-corder]

Thank you！I have updated the version of pytorch to 1.7. The issues have been solved.