2022 NICO Hybrid Context Generalization Challenge (ECCV 2022 Workshop) Forum

Go back to competition Back to thread list Post in this thread

> If we can use model ensemble or test-time augmentation?

As shown in Title

Posted by: vtddggg @ June 15, 2022, 6:09 a.m.

Multi-model ensembling is allowed as long as the whole model can be trained with 8 GPUs since only an 8-GPU server can be used for training the submitted model for each team in phase 2. If your models can be trained iteratively with 8 GPUs each time or they can be trained together with 8 GPUs, it meets the rule.
NICO challenge aims to evaluate the out-of-distribution / domain generalization ability instead of domain adaptation so that test-time fine-tuning is forbidden. Yet regular augmentation (such as random crop, flip, and colorjit) instead of model tuning in the test phase meets the rule.

Posted by: NICO-official @ June 20, 2022, 4:20 a.m.

Got it. Thanks for your response

Posted by: vtddggg @ June 21, 2022, 1:02 a.m.
Post in this thread