How is the coco_fewshot data distribution made?

[Caltech-Z]

Q1: How to make the part of seed1 to seed5?
Q2: Will the code work if I change to another type of dataset?
Thank you very much, I hope the owner can help me solve this problem.

[ZhangGongjie]

Q1: How to make the part of seed1 to seed5?
seed1 and seed5 are generated using random sampling with different random seeds. For each random seed, we ensure that there are exactly K object instances for each novel class. The numbers of base-class object instances vary.

Q2: Will the code work if I change to another type of dataset?
I think so, as long as you modify the other type of dataset into the same format.

[Caltech-Z]

Q1: How to make the part of seed1 to seed5? seed1 and seed5 are generated using random sampling with different random seeds. For each random seed, we ensure that there are exactly K object instances for each novel class. The numbers of base-class object instances vary.

Q2: Will the code work if I change to another type of dataset? I think so, as long as you modify the other type of dataset into the same format.

Thanks, I've taken care of that part.

[ZhaoPo]

Hi, if I want to change the MSCOCO data set to other types of data set, how to divide my own data set into coco_fewshot data format, can you provide the code of the data partition part? Looking forward to your reply very much.
[email protected]

[N-enthusiasm]

Q1: How to make the part of seed1 to seed5? seed1 and seed5 are generated using random sampling with different random seeds. For each random seed, we ensure that there are exactly K object instances for each novel class. The numbers of base-class object instances vary.
Q2: Will the code work if I change to another type of dataset? I think so, as long as you modify the other type of dataset into the same format.

Thanks, I've taken care of that part.

Could you please provide the code of the data partition part?