VISIONAR
A mobile iOS AR application using Unity AR Foundation and ARKit for real-time tracking, plane detection, object placement, and on-device object detection with YOLOv3-Tiny.
Overview
VisionAR is a mobile iOS augmented reality application built using Unity AR Foundation and ARKit. The app provides real-time device tracking, plane detection, object placement, and touch-based manipulation of virtual objects in physical environments.
AR Features
Tracking & Detection
The app leverages ARKit's capabilities for:
- Real-time device tracking — accurate 6DOF pose estimation
- Plane detection — horizontal and vertical surface detection for object placement
- Environmental understanding — mapping physical spaces for AR interaction
Touch Interactions
Touch-based interactions and Unity UI systems enable users to select, place, move, and manipulate virtual objects within detected physical environments. The interaction model is designed to feel intuitive and responsive on mobile.
Physics Integration
Rigidbody and collision systems support physically responsive virtual objects interacting with the AR environment. Placed objects respond to gravity and collisions, creating a more believable AR experience.
On-Device Object Detection
The app integrates Apple Vision and CoreML with a YOLOv3-Tiny model for on-device object detection. The model runs under mobile performance constraints, identifying real-world objects in the camera feed without requiring a network connection.
This enables the AR experience to be context-aware — understanding what objects exist in the user's environment and potentially responding to them.