YOLOv8n

Ultralytics YOLOv8n is the leanest model in the v8 lineup: it delivers high-speed object detection with 3.2 million parameters.

YOLOv8n (Nano) is the high-performance standard for edge-based computer vision. Developed by Ultralytics, this model achieves 37.3 mAP on the COCO dataset while maintaining a footprint of just 8.7 GFLOPs. It is designed for speed: frames process in roughly 0.99ms on an NVIDIA A100. This efficiency makes it the primary choice for real-time applications on constrained hardware (like the Raspberry Pi or Jetson Nano). The architecture supports multiple tasks including detection, segmentation, and pose estimation. It is the definitive tool for developers who need low-latency tracking without sacrificing accuracy.

https://github.com/ultralytics/ultralytics