What is enable_autoweighting parameter in the training spec file for detectnet-v2

Hello,

I’m training detectnet-v2 on custom dataset. The dataset contains 11 classes that are not balanced in the dataset. I assigned different weights (class_weight) to the classes in the training spec file to offset the imbalance. I’m not sure though whether to set the enable_autoweighing paramter to True or False.

Thank you

You can set it to True.

Refer to
https://docs.nvidia.com/metropolis/TLT/tlt-getting-started-guide/text/faqs.html

Distribute the dataset class: How do I balance the weight between classes if the dataset has significantly higher samples for one class versus another?

To account for imbalance, increase the class_weight for classes with fewer samples. You can also try disabling enable_autoweighting; in this case initial_weight is used to control cov/regression weighting. It is important to keep the number of samples of different classes balanced, which helps improve mAP.

Thank you