zeyang

I have the same concern with you and I have create an issue on the top. Maybe you can check that. I think I express the same idea with you.

> Hi there. > I well read uploaded source code. I found you set k-shot masked sample with k-'image' not 'instance'. > It might be okay that there is binary...

@lzhnb Would you also like to reproduce the result in COCO dataset or COCO to VOC setting? Maybe we can tackle it together. Here is my email [email protected]

Sorry. There is a small mistake with my email. The correct one is [email protected]. Thanks. Best, Ze ------------------ 原始邮件 ------------------ 发件人: "lzhnb"

Yes. This is actually due to lib import problem. You should not add lib. in front of the import directory, which makes it malfunction.

Actually, if you don't modify the original code, this problem won't arise. For me, at that time, I just modified _from xxx import zzz_ to _from lib.xxx import zzz_, for...

Could you reproduce the result with your latest code? I rerun your code for many times, but cannot get the result you indicate. Thanks.

Actually, r[k] will never be a list. Thus, I consider nothing will be updated through https://github.com/yanxp/MetaR-CNN/blob/353b4ed25e4b2ba23442f999b3990d5b90b30345/lib/roi_data_layer/roidb.py#L54.

> 1. I notice that in fine-tune stage, the COCO dataset only have one instance per image, despite of there may be number of potential instances, like some image contained...

> > Hi, In the Table 2 of the paper, you compare the average ap of mutiple runs of 10 seeds with the results given by TFA which is a...