Loading
Cookies help us deliver our services. By using our services, you agree to our use of cookies. Learn more

crowdAI is shutting down - please read our blog post for more information

Mapping Challenge

Building Missing Maps with Machine Learning


Completed
719
Submissions
1071
Participants
44974
Views

How to improve generalization

Posted by neptune.ml over 1 year ago

It seems that the models trained on competition dataset alone (as required) may not generalize to well due to the homogenous nature of the dataset.

We are thinking about retraining (or finetuning) the model with heavy augmentations and use scale/color TTA at inference.

What are your ideas?

1