Hello, I'm trying to use F-Clip on my own dataset. During aligning the anotations, I found the model input includes 'meta' (dict includes 'lpre' and 'lpre_label') from the york-urban dataset, while 'meta' was not mentioned in the paper and i couldn't find how it was used the loss calculations. So, would you mind explaining what is 'meta' and whether the model actually used it.
Sorry for the late reply, actually, "meta" of the input is just inherited from LCNN's code, you can remove it and also change the corresponding part of the code.
Hello, I'm trying to use F-Clip on my own dataset. During aligning the anotations, I found the model input includes 'meta' (dict includes 'lpre' and 'lpre_label') from the york-urban dataset, while 'meta' was not mentioned in the paper and i couldn't find how it was used the loss calculations. So, would you mind explaining what is 'meta' and whether the model actually used it.