Digital Eyes on the Road: Using Street Cameras to Verify Traffic Integrity and...
summary
An end-to-end pipeline links NYC DOT street cameras (200+ in Manhattan) to TomTom road-level telemetry, uses low-resolution YOLO-based object counts, aligns and smooths time series per camera-to-road match, and flags discrepancies with a residual-based detector that compares camera-observed vehicle counts to GPS-derived intensity. The system detects injected Sybil slowdowns with 91.5% detection at Δ=+0.05 and ≅99% for stronger attacks, and preserves coverage via neighbor consensus at 96.68% of tested Manhattan locations.
AI-written summary · about
chapters (8)
- 0:00Traffic integrity threats
- 3:49GPS-based defenses limits
- 5:14Street cameras for validation
- 8:04Matching camera and GPS data
- 11:59Detecting attacks with residuals
- 13:49Neighbor consensus gaps
- 14:53City-wide evaluation results
- 15:57Future directions
related talks
24mBeyond the Frame: Using Computer Vision to Unlock the Potential of Video Data
46mML Use Cases at Think Big Analytics
01h 06mSelf-Driving and Connected Cars: Fooling Sensors and Tracking Drivers
19mHow we leverage AI to build great maps at massive scale
22mUrban Artificial Intelligence in Amsterdam
32mData&AI in TomTom: How are are transforming mobility
38mFinding Lane Lines for Self Driving Cars
28mEstimating Traffic Flow to find New Drive Thru Locations