DarkNetV2

The DarkNetV2 model is based on the “YOLO9000: Better, Faster, Stronger” paper.

Architecture overview

This paper improves its version version by adding more recent gradient flow facilitators.

The key takeaways from the paper are the following:

  • adds batch normalization layers compared to DarkNetV1

Model builders

The following model builders can be used to instantiate a DarknetV2 model, with or without pre-trained weights. All the model builders internally rely on the holocron.models.classification.darknetv2.DarknetV2 base class. Please refer to the source code for more details about this class.

darknet19([pretrained, checkpoint, progress])

Darknet-19 from "YOLO9000: Better, Faster, Stronger"