YOLOv8 Pose

Another feature provided by YOLOv8 is pose estimation. For this task, YOLOv8 was pretrained on the COCO dataset. The model outputs 17 2D keypoints with an mAP50 of 90.2, corresponding to mean Average Precision at a 50% IoU threshold. The model runs in real-time and accurately estimates the pose even in crowd scenes.

YOLOv8 Pose

Example

from dronevis.models import YOLOv8Pose

model = YOLOv8Pose()
model.load_model()
model.detect_webcam()

YOLOv8 Pose Class

class dronevis.models.YOLOv8Pose(track=False, show_conf=True, show_labels=True)

YOLOv8 model implementation for pose estimation

load_model(model_weights='yolov8n-pose.pt')

Load model weights

Parameters

  • model_weights (str, optional) – Path to model weight or the name of the official

  • in the ultralytics website which will be downloaded automatically. (weights) –

  • to "yolov8-pose.pt". (Defaults) –