
Case brief
Autonomous Vehicle Startup Needs Edge-Case Driving Data
Customer
Self-driving car startup (20 employees)
- 01Problem
- Their lane detection model fails on faded markings and wet roads. Public datasets mostly contain clean, well-marked California highways. They need 500 images of worn lane markings at intersections in snowy cities.
- 02Solution
- They post a request on DataW: "$150 for 50 intersection photos with faded lane markings in Chicago, Boston, or Toronto." Collectors upload photos within 3 days.
- 03Outcome
- Model improves 22% on edge cases. Cost: $150 vs. $5,000+ from enterprise vendors. Time saved: 3 days vs. 6 weeks.

















