Fig. 8
From: Towards autonomous robotic THz-based in vivo skin sensing: the PicoBot

Marker detection and pose estimation (a) 4 green-colored marker detection on the cheeks region using Yolo-V5 detection for point cloud segmentation and surface normal estimation, (b) Convex hull generated from the markers and defining the center point to estimate the pose at the point, (c) Pose estimation and aligning target orientation with end-effector tip coordinate frame.