Scanning & AR
Room Scanning AI AppUnder NDA
An iOS app for spatial intelligence scanning of indoor environments. Uses iPhone LiDAR and computer vision to automatically detect architectural elements — windows and doors — while the user scans a room, enabling accurate digital reconstruction of interior spaces.
Scope of work
- Implemented room scanning flow using ARKit and LiDAR
- Implemented automatic window and door detection during room scan
- Implemented real-time overlay of detected architectural elements on scan view
- Integrated Vision framework for image-based element recognition
- Integrated CoreML model for architectural feature classification
- Implemented accuracy improvement pipeline for detection results
Challenges & solutions
The primary challenge was detection accuracy — window and door boundaries are difficult to reliably identify from LiDAR point cloud data alone due to reflective surfaces, varying lighting conditions, and thin structural edges. Improved accuracy by combining LiDAR depth data with Vision framework image analysis, using multi-frame averaging to reduce false positives and applying confidence thresholds to filter unreliable detections before presenting results to the user.