This project will develop a roadmap for a DDOT-wide approach to automated image processing that will ultimately enhance DDOT’s ability to assess traveler behavior and roadway conditions for planning, design, and operations.
DDOT frequently uses camera imagery (e.g., panoramic street-level photography, time-lapse cameras, closed-circuit television (CCTV) cameras) to better understand traveler behavior and existing configuration and condition of the roadway and associated infrastructure. This imagery can provide insight into:
Existing patterns or conditions for future planning or design
Real-time conditions for operational decision making
Impacts of a change in condition (i.e., before-and-after analyses)
To date, much of the processing of this imagery has been done manually, which often proves costly and inefficient, thereby limiting the degree to which DDOT is able to use camera imagery.
Recent advances in artificial intelligence and machine learning have the potential to speed up and improve processes for analyzing camera imagery. DDOT staff have engaged with vendors to explore some of these possibilities on a limited scale, but have recognized the need for a consistent, agency-wide approach to ensure quality of analysis, maximize utility across divisions, and minimize any duplication of effort.
Desired Outcome & Expected Benefits
This project will provide comprehensive recommendations on how DDOT can expand its use of camera footage via automated image processing to ensure all agency information needs are met. Results will directly identify actions and changes that DDOT can make to existing technology, policy, and processes to ensure quality of analysis, maximize utility across divisions, and minimize any duplication of effort. The project will benefit the District by enhancing DDOT’s ability to understand traveler behavior and roadway conditions toward better planning, design, and operational decision-making.
The primary objective of this project is to build up DDOT’s automated image processing capability. This will be achieved through a series of tasks:
Assessment of use cases and needs via internal stakeholder outreach
Market research on available tools and technologies, including use cases, implementations to date, required supportive technologies and processes, and validation and verification
Identification of existing gaps in supportive technologies and processes (e.g., installation and data storage)
Comprehensive recommendations on next steps for standing up a consolidated program that can effectively and efficiently support need for automated image processing agency wide.
Potential use cases include, but are not limited to: vehicle trajectories; parking occupancy detection; vehicle, freight, cyclist, pedestrian counts and turning movements; collision and near miss detection; pavement and sidewalk condition; lane and signal configuration; sign and pole inventory.