Makes one wonder if the YOLO algorithm would work better with hexagons.

"Hexagons were an important choice because people in a city are often in motion, and hexagons minimize the quantization error introduced when users move through a city. Hexagons also allow us to approximate radiuses easily, such as in this example using Elasticsearch."

[Edit]Maybe https://www.researchgate.net/publication/372766828_YOLOv8_fo...