Using AR and YOLOv8-Based Object Detection to Support Real-World Visual Search in Industrial Workshop: Lessons Learned from a Pilot Study.

Visual search requires focused attention to distinguish the target object from its surroundings. This task is increasingly more difficult when searching for a given object within a messy industrial workshop. Thus, we explored how we can support visual search with the help of an augmented reality (AR) head-mounted display (HMD) running YOLOv8 object detection model. We report on the results from a pilot user study with eleven participants tasked with searching for a series of objects in a real-world workshop. Consequently, we reflect on lessons learned concerning the experimental setup with AR interface enhanced with Segment Anything Model (SAM) for YOLOv8 training.