Inspiration
Measuring and visualizing room layouts has always been time-consuming and imprecise. Interior designers, homeowners, and renovation teams often struggle to understand a space before making decisions. We wanted to eliminate tape measures, guesswork, and manual modeling — transforming the simple act of scanning a wall into an intelligent, real-time digital reconstruction.
What it does
RoomScannerAI uses a smartphone or tablet camera to scan indoor environments, detect walls, and automatically build an accurate 3D room model. The system recognizes key wall elements like doors, windows, and power outlets, and places them correctly in the virtual layout. Users can also apply different textures and finishes to walls, choose room shapes, or upload photos for reconstruction — allowing them to visualize design changes instantly.
How we built it
We combined ARCore/ARKit plane detection with a custom computer vision AI pipeline for real-time object recognition. A geometry engine tracks wall surfaces and stitches them together to infer room corners and structure. We developed a rendering layer to apply materials and provide live visual feedback to the user. The app was built with scalable architecture so photo-based scanning and design features can seamlessly integrate with AR scanning.
Challenges we ran into
Handling sensor noise and ensuring accurate wall stitching in uneven environments Achieving reliable object detection on mobile hardware in real-time Aligning AR coordinate planes into a clean, watertight 3D room model Designing an intuitive user experience while scanning moving around the room Optimizing texture rendering to maintain smooth performance Accomplishments that we're proud of Delivered a functional prototype capable of real-time wall detection and 3D model generation Successfully recognized and placed real-world elements like windows and doo Implemented interactive design features such as texture/material overlays Created a flexible, multi-input capture system: live scan, camera photos, or uploads Built a strong foundation for a full-scale commercial-ready solution
What we learned
We deepened our expertise in AR frameworks, 3D geometry reconstruction, and mobile AI deployment. We also learned how critical user guidance and feedback loops are when working with spatial scanning. Most importantly, we gained insight into bridging real-world environments with digital design tools using computer vision.
What's next for RoomScannerAI
Enhanced accuracy using LiDAR devices and depth fusion Support for additional room shapes and multi-room layouts More detailed object detection (switchboards, vents, furnishings) Export to CAD/BIM formats for professionals (e.g., AutoCAD, Revit) Cloud sync and collaborative design features Integration with AR interior design and shopping tools
Log in or sign up for Devpost to join the conversation.