A Vision-Based Analysis of Congestion Pricing in New York City

TL;DR

This work addresses the challenge of evaluating congestion pricing in a major city by analyzing traffic patterns from a network of public camera feeds. It introduces a vision-based pipeline, including a low-resolution detector named YOLO-LR, a scalable distributed processing stack, and a bias-aware pattern-analysis framework to quantify density changes around the Congestion Relief Zone. Key contributions include optimizing detection for $352\times240$ frames, enabling real-time processing of hundreds of feeds on GPUs via TensorRT, and a congestion-metrics approach based on Peak Hour Differential, defined as $\text{PHD}_{s,w} = \text{Peak}_{s,w,\text{after}} - \text{Peak}_{s,w,\text{before}}$, utilizing rolling means $\bar{D}_s(t)$ and conditioned expectations $\mu_{s,h,w,p}$. The approach supports quantitative policy evaluation with implications for urban mobility and smart-city design.

Abstract

We examine the impact of New York City's congestion pricing program through automated analysis of traffic camera data. Our computer vision pipeline processes footage from over 900 cameras distributed throughout Manhattan and New York, comparing traffic patterns from November 2024 through the program's implementation in January 2025 until January 2026. We establish baseline traffic patterns and identify systematic changes in vehicle density across the monitored region.