Numpy error?

[seanmmorais]

Hi all, trying to run train.py for coseg_seg, but keep running into this error:

C:\Users\Smorais\Anaconda3\envs\meshcnn\lib\site-packages\torch\package_directory_reader.py:17: UserWarning: Failed to initialize NumPy: No module named 'numpy.core._multiarray_umath' (Triggered internally at ..\torch\csrc\utils\tensor_numpy.cpp:68.)
dtype_to_storage = {data_type(0).dtype: data_type for data_type in storages}
------------ Options -------------
arch: meshunet
batch_size: 12
beta1: 0.9
checkpoints_dir: ./checkpoints
continue_train: False
dataroot: datasets/coseg_aliens
dataset_mode: segmentation
epoch_count: 1
export_folder:
fc_n: 100
flip_edges: 0
gpu_ids: []
init_gain: 0.02
init_type: normal
is_train: True
lr: 0.001
lr_decay_iters: 50
lr_policy: lambda
max_dataset_size: inf
name: coseg_aliens
ncf: [32, 64, 128, 256]
ninput_edges: 2280
niter: 100
niter_decay: 500
no_vis: False
norm: batch
num_aug: 20
num_groups: 16
num_threads: 3
phase: train
pool_res: [1800, 1350, 600]
print_freq: 10
resblocks: 3
run_test_freq: 1
save_epoch_freq: 1
save_latest_freq: 250
scale_verts: False
seed: None
serial_batches: False
slide_verts: 0.2
verbose_plot: False
which_epoch: latest
-------------- End ----------------
computing mean std from train data...
Traceback (most recent call last):
File "train.py", line 10, in
dataset = DataLoader(opt)
File "C:\Users\Smorais\Documents\MeshCNN\data_init.py", line 21, in init
self.dataset = CreateDataset(opt)
File "C:\Users\Smorais\Documents\MeshCNN\data_init.py", line 9, in CreateDataset
dataset = SegmentationData(opt)
File "C:\Users\Smorais\Documents\MeshCNN\data\segmentation_data.py", line 22, in init
self.get_mean_std()
File "C:\Users\Smorais\Documents\MeshCNN\data\base_dataset.py", line 32, in get_mean_std
for i, data in enumerate(self):
File "C:\Users\Smorais\Documents\MeshCNN\data\segmentation_data.py", line 29, in getitem
mesh = Mesh(file=path, opt=self.opt, hold_history=True, export_folder=self.opt.export_folder)
File "C:\Users\Smorais\Documents\MeshCNN\models\layers\mesh.py", line 16, in init
fill_mesh(self, file, opt)
File "C:\Users\Smorais\Documents\MeshCNN\models\layers\mesh_prepare.py", line 21, in fill_mesh
mesh2fill.ve = mesh_data['ve']
File "C:\Users\Smorais\Anaconda3\envs\meshcnn\lib\site-packages\numpy\lib\npyio.py", line 255, in getitem
raise KeyError("%s is not a file in the archive" % key)
KeyError: 've is not a file in the archive'

This is my command:
python train.py --dataroot datasets/coseg_aliens --name coseg_aliens --arch meshunet --dataset_mode segmentation --ncf 32 64 128 256 --ninput_edges 2280 --pool_res 1800 1350 600 --resblocks 3 --lr 0.001 --batch_size 12 --num_aug 20 --slide_verts 0.2

Anyone have any idea what's going on? I've tried different versions of numpy (1.15.0, 1.19) and get the same error.

[wang-hairui]

I guess that's a problem with the version update. Have you tried downgrading the numpy?
pip install numpy==1.21.6

[Broadcloak]

My Python version is 3.6.8, I reinstalled numpy 1.19.2 and it worked!