YOLOv8-p2 COCO Pretrained Model

This model is a YOLOv8-p2 model trained on the COCO dataset, with P2-P5 output layers.

Example Usage

from huggingface_hub import hf_hub_download
from ultralytics import YOLO
from PIL import Image

model_path = hf_hub_download("soyeollee/yolov8x-p2-coco", "model.pt")
model = YOLO(model_path)

image_path = "/path/to/image"
output = model(Image.open(image_path))

Performance

metric (maxDets=100)	yolov8x (official)	yolov8x-p2 (this repo)	yolov8x-p26	yolov11x (official)	yolo11x-p2
AP @[ IoU=0.50:0.95 / area= all ]	0.540	0.541	0.544	0.546	0.553
AP @[ IoU=0.50 / area= all ]	0.710	0.712	0.713	0.716	0.722
AP @[ IoU=0.75 / area= all ]	0.588	0.590	0.593	0.595	0.606
AP @[ IoU=0.50:0.95 / area= small ]	0.360	0.386	0.381	0.377	0.386
AP @[ IoU=0.50:0.95 / area=medium ]	0.594	0.586	0.586	0.597	0.595
AP @[ IoU=0.50:0.95 / area= large ]	0.707	0.686	0.691	0.702	0.702