na-z / sdcot Goto Github PK

Hello, Professor Zhao
Thank you for your interesting work.
However, I have some problems with the splitting of the dataset SUNRGBD.
Can I only use 3203 training data in the case of 5 classes data in the base stage and 2492 training data in the case of 5 classes data in the novel stage? The same situation happened in the other two Settings ( 3234 data for 7 base stage training, 2178 data for 3 novel stage training / 3826 data for 9 base stage training, 168 data for 1 novel stage training ).
The reason I raise this question is that it seems difficult to train with only 168 training data.
If there is any problem with my code understanding, please tell me.
Best regards !