Upgrade to medium size model

WATonomous / wato_monorepo

Dockerized ROS2 stack for the WATonomous Autonomous Driving Software Pipeline

Apache License 2.0

89 stars 1 forks source link

Closed Gongsta closed 8 months ago

Gongsta commented 9 months ago

Better accuracy. We can either use YOLOv8m or YOLOv8l (YOLOv8x is overkill). Settling for medium as it seems a good compromise.

Also increased default model resolution

Gongsta commented 8 months ago

vram is ~2x more used per model