Scaled-YOLOv4 model

In the previous article, we have discussed about YOLOv4's architecture, and how it became a SOTA (state-of-the-art) model for the object detection task offering the best performance in terms of both speed and accuracy. But it could not beat Google's EfficientDet in terms of overall accuracy on the COCO dataset.

So, the authors of YOLOv4 came back and pushed the YOLOv4 model forward by scaling it's design and scale and thus outperforming the benchmarks of EfficientDet. This resulted in Scaled-YOL...