AttributeError: 'NoneType' object has no attribute 'shape'

[fky07]

Why is running 2_ Gen_ Support_ There was an issue with the pool.py file:AttributeError: 'NoneType' object has no attribute 'shape'
Is it because there are grayscale images？

[GuangxingHan]

Grayscale images should have valid shape. Can you make sure that you have the correct path and get the right image tensor?

[fky07]

Grayscale images should have valid shape. Can you make sure that you have the correct path and get the right image tensor?

Thank you, there is a problem with the format of the picture I downloaded, it has been solved now.
But : due to I didn't find how to implement these two steps in the code，I would like to ask: In the second and third steps of meta-training, Is the corresponding code in meta_faster_rcnn/modeling/fsod?

[GuangxingHan]

Yes, they are all implemented in the codebase.

[fky07]

thank you for your reply

[fky07]

Excuse me again, I don't understand a lot: In the stage of meta-training: BASE in .yaml: "Base-FSOD-C4.yaml" is the object detection model used for meta-training? I don't know its specific role in the meta-training process, can you tell me?