AndreaMacri-sys
commented
2 years ago SOLVED backtracing the error I i found this function
def get_classes(cls, classes=None):
"""Get class names of current dataset.
Args:
classes (Sequence[str] | str | None): If classes is None, use
default CLASSES defined by builtin dataset. If classes is a
string, take it as a file name. The file contains the name of
classes where each line contains one class name. If classes is
a tuple or list, override the CLASSES defined by the dataset.
Returns:
tuple[str] or list[str]: Names of categories of the dataset.
"""
if classes is None:
return cls.CLASSES
if isinstance(classes, str):
# take it as a file path
class_names = mmcv.list_from_file(classes)
elif isinstance(classes, (tuple, list)):
class_names = classes
else:
raise ValueError(f'Unsupported type {type(classes)} of classes.')
return class_namesin mmdet/datasets/custom.py, which explains everything.
So to avoid the error just replace classes = ('tube') with classes = ('tube', ) or classes = ['tube'] in the data_instance.py.
Sorry for opening the issue, hope it can be at least helpful for someone else in the future.
OranRonen94
Premising that I have already used mmdet several time with different custom datasets and/or configurations, without any problem. This time my dataset is composed by one only class (tube), so I prepared the config/base/dataset/my_data_instance.py and and the configs/base/model/mask_rcnn_r50_fpn.py. Now when I launch python tools/train.py [config_path] I got this strange error
I tried to add a dummy second class (and set num_class to 2) and the error disappears. What is it due to? And how can I solve this to work with one class only?
Configs
data_instance
model
Environment