As someone interested in the intersection of interfaces and mixed reality, ObjectDetector is a project which merges the Meta YOLOv9 Sentis sample with the OpenXR microgestures interaction to create an easy way to visually filter objects in your surrounding environment, inspired by Anduril's EagleEye helmet interface and the Kiroshi scanner from Cyberpunk 2077.

A user can use thumb swipes to navigate a HUD and select toggles which either show all detections, hide all detections, or hide/show specific classes from the 80 trained inputs of the YOLOv9 AI model.

Although this is mostly a proof of concept, in the future this could easily utilise more specialised vision models and with better hardware to achieve higher accuracy performance!