PocketInspector
A sub-$100 USB-C clip-on camera that runs on-device crack, gap, and edge detection for tradespeople who need quick visual records and measurements on-site.
Independent tradespeople (carpenters, tile installers, drywall finishers) doing on-site quality checks
- Plug-in USB-C camera with on-device inference (no cloud, no lag)
- Tap-to-detect cracks, gaps, and straight edges with calibrated measurements
- Photo log with timestamps, GPS, and voice-note annotations
- One-tap PDF job report export for client and warranty records
MiLSD-class tiny models now fit on cheap MCUs, finally making once-enterprise inspection tools affordable for a single tradesperson's toolkit.
Inspection cameras are a large, mature category (40,000+ Amazon results, sold at Home Depot/Canadian Tire) and the digital inspection market is ~$21B growing 4-7% CAGR, but most are borescopes/endoscopes for pipes/walls rather than clip-on surface crack/gap/edge tools for finish trades.Amazon.ca: Inspection Cameras ↗Digital Inspection Market Size, Share & Forecast to 2030 ↗
AI crack/defect detection is dominated by enterprise vendors (iFactory, Cyfersoft, Global Gauge) and DIY hobbyist builds (CrackScan on Raspberry Pi); the specific niche of a sub-$100 USB-C clip-on with on-device ML aimed at individual carpenters/tile/drywall tradespeople is largely unoccupied.AI Vision Crack & Structural Defect Detection ↗GitHub - gsai21/CrackScan ↗Edge AI USB Camera for Computer Vision Applications ↗
Comparable USB inspection cameras retail $30-$150 and tradespeople do buy premium tools, but at sub-$100 with a dedicated ML chip and custom-trained models margins are thin; free/cheap smartphone crack-detection apps create a strong substitute that suppresses willingness to pay for a dedicated hardware device.INSPECTION CAMERAS - The Home Depot ↗Digital Tape Measure | Walmart Canada ↗
On-device tiny-model AI on MCUs is a durable trend and finish-trade quality checks won't disappear, but the clip-on USB-C camera form factor faces medium-term disruption from ever-better smartphone cameras plus on-device AI apps, capping long-term defensibility.Digital Inspection Market Size & Share [2026-2035] ↗
Hardware is buildable: edge-AI USB cameras (e-con EdgeECAM50) and MiLSD-class line-segment models on Cortex-M class MCUs are proven, and CrackScan shows similar pipelines running on Raspberry Pi; main cost/risk is collecting labelled training data for trade-specific cracks, gaps and edges.Edge AI USB Camera for Computer Vision Applications ↗GitHub - gsai21/CrackScan ↗