AI Detection of Trash with Automated Enforcement for a Cleaner and Smarter City | NYU Tandon School of Engineering

AI Detection of Trash with Automated Enforcement for a Cleaner and Smarter City

Center for Urban Science + Progress
Sustainability & Environment,
Urban
Project Sponsor:

Edward Boze, Chief Innovation Officer, Office of the Mayor, City of Paterson

 

MENTOR:

Bandana Parajuli, Data Analyst, Office of Innovation, City of Paterson

Authors

Suyan Ying, Guangyuan Yang, Zhi Jie Wang

Research Question

How can AI be integrated with mobile camera systems on city vehicles to detect illegal dumping and support automated enforcement processes? This project explores how technology can improve sanitation outcomes, reduce manual inspection efforts, and foster greater resident accountability in waste management practices.

Background

The City of Paterson is a municipal government located in Passaic County, New Jersey. Established in 1791 and incorporated in 1851, it is the third-largest city in the state with a population of approximately 160,000. Paterson operates under the Faulkner Act (Mayor-Council system), with a Mayor serving as the Chief Executive Officer and a City Council providing legislative oversight.

Pervasive illegal dumping and trash on the streets and sidewalks is a challenge in the City of Paterson, undermining neighborhood quality, also creating an "anything goes” mentality in the city, causing further degradation in the quality of life. Public Works staff members are unable to adequately enforce trash ordinances manually, and the resultant success of prosecutions is in question. Mobile cameras exist on street-sweeper vehicles that navigate each city street once per week, but are only used for prosecution and not detection.

Methodology

The project involves:

  1. Using AI to analyze the camera footage to detect illegal dumping and trash.

  2. Associating that infraction with a property owner likely via AI, geospatial tagging, and property data.

  3. Automating the issuance of citations via a bot.

  4. Tracking the successful prosecution of citations. Some manual intervention is allowed in the solution.

The solution may involve identifying existing commercial products or delivering a novel solution, in whole or in part.

The students are working with the Innovation Team to develop a proof-of-concept. AI tasks may include exploring object detection models (e.g., identifying trash left outside designated pickup days), testing machine learning approaches.

The goal is to determine the degree to which the process of detection and enforcement can be automated, promoting cleaner streets and increasing resident accountability.

Deliverables
  • AI-Based Detection Prototype: A working computer vision model capable of identifying illegal dumping from sample mobile camera footage.
  • Operational Workflow Design: A proposed system for linking AI detections to municipal enforcement processes, including citation generation.
  • Interactive Dashboard or Visual Report: A data visualization tool or dashboard highlighting dumping hotspots, detection results, and potential impact on sanitation operations.
Data Sources

City of Paterson is providing the following data:

  • Camera Footage: Sample video recordings from cameras mounted on street sweeper vehicles are provided to support the development of the AI detection model.

  • Service Request Data: Historical records of illegal dumping and sanitation-related complaints from the city's 311 platform (e.g., SeeClickFix).
     
  • Geospatial Data: GIS layers including property parcels, zoning boundaries, ward boundaries, and street maps will be provided by the City's GIS team.
     
  • Enforcement and Citation Records: Anonymized data on past code violations or tickets related to illegal dumping, if available, are shared to inform workflow design and enforcement analysis.