AssertionError: Attribute 'thing_classes' in the metadata of 'objects365_train' cannot be set to a different value!

[LigZhong]

📚 Documentation

I arrange the dataset as guided in the documentation, but get this error when trying to train split detectors.

• Links to the relevant documentation/comment:

[xingyizhou]

Hi,
Sorry for the delayed response. Can you specify your command?

[ahong007007]

I have the same problem.

Partitioned detector

python projects/UniDet/train_net.py --config-file projects/UniDet/configs/Partitioned_COI_R50_2x.yaml --num-gpus 2

[JR-Wang]

I have the same problem.
Have you solved this problem? It seems 'catetgories' defined in 'unidet/data/datasets/objects365.py' is different from both annotations of object365 2019 version and 2020 version.
@xingyizhou Would you please share the object365 dataset you used? Many thanks.
Partitioned detector

python projects/UniDet/train_net.py --config-file projects/UniDet/configs/Partitioned_COI_R50_2x.yaml --num-gpus 2

[322jtZhang]

I have the same problem.
Have you solved this problem? It seems 'catetgories' defined in 'unidet/data/datasets/objects365.py' is different from both annotations of object365 2019 version and 2020 version.
@xingyizhou Would you please share the object365 dataset you used? Many thanks.
Partitioned detector
python projects/UniDet/train_net.py --config-file projects/UniDet/configs/Partitioned_COI_R50_2x.yaml --num-gpus 2

I have the same question, too. How to solve this? thanks