How to Train a Scaled-YOLOv4 Object Detection Model

This is neat, but it's less "How to Train a YOLO4" and "How to put your data into [our product that abstracts the training of YOLO4 behind the scenes so you don't even have to know what you're doing]." That's totally fine, to be clear, but the title makes it seem like one will learn about training YOLO4 models, when instead one is really learning how to use the roboflow dashboard to put images into their system, the fact that it happens to be YOLO4 under the hood is incidental.

This breakdown of the internals may be more what you were hoping for: https://blog.roboflow.com/a-thorough-breakdown-of-yolov4/

Very cool seeing the YOLO v4 continued to be used after all of the naming fiasco.

These results are surprisingly good.

To complicate matters further, the code for this improvement of YOLOv4 (published by one of the original YOLOv4 authors) is actually a fork of the YOLOv5 repo: https://github.com/WongKinYiu/ScaledYOLOv4/tree/yolov4-csp

I always see these kind of things for Object detection in 2D images. What about for example feature detection in 3D meshes/ DICOM scan data? Are there equivalents to the 2D models?

Yep: https://github.com/XuyangBai/D3Feat