GarageLens
A pocket assistant for DIY car and motorcycle owners that points the camera at a part under the hood, identifies it, and overlays torque specs, common failure signs, and the right YouTube tutorial.
Weekend DIY car and motorcycle owners
- On-device segmentation that isolates bolts, hoses, and connectors from the cluttered engine bay
- Live overlay of torque values, fluid capacities, and part numbers when you tap an object
- Fault-mode hints ('this hose cracks at 80k km') drawn from owner-forum knowledge
- Save-the-car memory so you can re-scan the same bay six months later
Lightweight vision foundation models are now efficient enough to run semantic segmentation and depth on a phone, finally enabling real-time AR guidance for messy real-world scenes like an engine bay.
DIY auto market grew ~65% from 2017-2025 per Auto Care Association data, and several recent AI car-scanner apps have launched, indicating active user interest.DIY Auto Maintenance Market Size, Growth & Research Report ↗CarAI Scan | AI Car Scanner App with AR Guidance & Phone Camera ↗
Direct competitors already exist for nearly every sub-feature: CarAI Scan (AR guidance), MechanicML, Partivize, CarSight, Car Part Identifier by Image for cars; Torque & Horse, MotoStaq, and TorqueAI explicitly cover motorcycles + torque specs — the proposed differentiators (AR overlay, torque specs, motorcycles) are already being shipped.CarAI Scan | AI Car Scanner App with AR Guidance & Phone Camera ↗Torque & Horse — Your AI Garage for Motorcycles ↗MechanicML | AI Car Part Identification ↗
Competitors monetize via auto-renewing subscriptions (e.g., Car Part Identifier on App Store) and freemium; the DIY auto market is large enough to support paid tools, but willingness-to-pay is moderate given many free alternatives and YouTube availability.Car Part Identifier App - App Store ↗DIY Auto Maintenance Market Size, Growth & Research Report ↗
ICE/petrol DIY maintenance will persist for decades given 1.4B+ vehicles on the road and rising repair costs pushing owners to DIY; EV shift adds uncertainty but motorcycles and older cars keep the base stable.
Building real-time part segmentation + AR overlay on cluttered engine bays is non-trivial — small parts, glare, oil, occlusion, and depth estimation still struggle; feasible as a photo-snap + LLM lookup but real-time AR guidance is harder than the source paper's benchmarks suggest.