Data Management and Computer Vision

Notice of Intent to Sole Source University of Utah, 270 S 1500 E Salt Lake City, UT 84112 has a requirement for Collaborative research project between the Tissue Injury Branch (TIB) and the University of Utah to perform a variety of data management and computer vision tasks that support the overall implementation of the Collision Vision Research Project.

This notice is being posted so that any unknown vendors who are capable of providing the support that may come forward and express their interest and capability in providing implementation of the Collision Vision Research Project. If your company/organization is capable of providing the required item, provide your entities capability statement to Anita Edwards, Contract Specialist by 11/18/2024 4:00 PM EST via email to anita.edwards@nih.gov.

In addition to your capability statement also provide any applicable distribution agreements regarding Collision Vision Research Project, the size and socioeconomic status of your company, and any other information pertinent to this requirement. Reference Procurement ID number NIHNR2401021 in the subject of the email.

This is not a request for quote, no quotes should be sent in response to this notice.

Statement of Work: Collaborative research project between TIB and the University of Utah to identify the relationship between built environment characteristics and global vehicle collision risk. Specifically, we will leverage Google Street View images and use advanced neural networks to extract relevant built environment characteristics from street images. We will use these data to quantify associations between built environment characteristics and global collision risk. The specific study aims of Collision Vision are as follows:

• Aim 1. Develop computer vision techniques to produce vehicle collision risk indicators. Create areal-level indicators of pedestrian risk (e.g., crosswalks, sidewalks, streetlights) and bicyclist risk (e.g., bike lanes), and motor risk (e.g., number of lanes), from Google Street View images combined with other area measures such as sociodemographic and motorization. • Aim 2. Measure the accuracy of data algorithms and construct an interactive geoportal.

Assess the accuracy of computer vision algorithms by comparing against manual annotations (produced via crowdsourcing and the research team). Algorithms will be modified in an iterative refinement process to maximize accuracy. Construct an interactive geoportal for data visualization and data sharing. • Aim 3. Utilize our global repository, Collison Vision, and a large collection of injury and fatality records to evaluate built environment impacts on motor vehicle collision risk.

Hypotheses: Places with street designs allowing protected movement of pedestrians and bicyclists and speed dampening characteristics will have lower risk of injury and mortality. Collaborators at the University of Utah will provide data management and development of computer vision models to extract global built environment features. The project deliverables include creating the training image dataset, computer vision model building, and global image analysis.

Background: NINR’s mission is to lead nursing research to solve pressing health challenges and inform practice and policy—achieving better health for everyone. Nursing research views people from multiple perspectives—from biological factors to the society in which they live. By bringing these perspectives together, nursing research seeks to understand and address a holistic picture of health. One of the pressing issues of our time are motor vehicle collisions.

Each year 1.35 million people are killed on roadways around the world. Globally, crash injuries are the eighth leading cause of death for all age groups, and they are the top leading cause of death for young people aged 5-29 years of age. The United States also has higher rates of fatal vehicle collisions than most other high-income countries.

Reduction in these crash rates would have powerful societal level impacts by protecting our young people, enabling them to contribute economically, politically, and socially to their communities. Fatal and nonfatal collisions will cost the global economy $1.8 trillion between 2015-2030. The Sustainable Development Goal (SDG) on road traffic deaths is to half of them by 2030, but few countries are moving in the right direction towards meeting that goal.

To provide this much needed data, our project will produce, for the first time, a global data repository that that provides roadway crash factors and risk level. We will leverage Google Street View images across the world and advanced neural networks to extract relevant built environment characteristics from street images. We will use these data to quantify associations between built environment characteristics and global collision risk.