← TrendWatcher
arXiv cs.AI
6/10

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.

Target user

Weekend DIY car and motorcycle owners

Features
  • 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
Why now

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.

Signals · overall 6/10
Demand
7/10

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 ReportCarAI Scan | AI Car Scanner App with AR Guidance & Phone Camera

Whitespace
3/10

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 CameraTorque & Horse — Your AI Garage for MotorcyclesMechanicML | AI Car Part Identification

Monetization
6/10

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 StoreDIY Auto Maintenance Market Size, Growth & Research Report

Longevity
7/10

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.

Feasibility
5/10

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.

DPNeXt: A Lightweight Multi-Scale Feature Fusion Framework for Efficient ViT-Based Multi-Task Dense PredictionarXiv cs.AI · 2026-07-20 (4d ago)