PSCR Launches Challenge to Help with Computer Image Quality
Friday, August 28, 2020 | Comments

The Public Safety Communications Research (PSCR) division of the National Institute of Standards and Technology (NIST) launched the Enhancing Computer Vision for Public Safety challenge to create a new line of research into computer vision that will assist with the development of life-saving tools for public safety. The submission period for the first phase of the challenge opens on September 8.

Public-safety officers who rely on computer vision in the field are often limited by image quality issues such as dirt or grease on a camera lens, poor lighting or simply low-quality camera equipment. To help first responders, PSCR wants to understand what "good quality" means to a computer vision algorithm.

The prize challenge has three goals:
• Create training data, including images and videos depicting camera impairments that hinder computer vision algorithms
• Measure failure rate and brainstorm the best methods to assess the likelihood that the computer vision algorithms can make reliable decisions
• Open data to inspire new research

This will enable a new line of research, into metrics that will compliment computer vision. Referred to as no reference (NR) metrics for computer vision, these algorithms will automatically identify quality problems that hinder computer vision algorithms. PSCR envisions using computer vision systems in combination with NR metrics to mitigate camera impairments. For example, when noise is detected, choose a computer vision algorithm that is less accurate overall but has robust performance on noisy images. By making these computer vision systems smarter, emergency operations can become safer.

PSCR, along with challenge partner First Responder Network Authority (FirstNet Authority), invites contestants to submit a concept paper proposing a dataset focused on one type of camera impairment and outlining the best method to assess computer vision failure rates for those media.

Winning concept papers will be invited to compete in phase 2 where contestant teams will implement these experimental designs and help further computer vision research. Teams or individuals who enter the contest will compete for up to a total of $28,000 each while contributing research for America's first responders.

Find more information about the challenge here. Concept papers are due by 5 p.m. MT Tuesday, Oct. 20. Would you like to comment on this story? Find our comments system below.

Post a comment
Name: *
Email: *
Title: *
Comment: *


No Comments Submitted Yet

Be the first by using the form above to submit a comment!

Magazines in Print

September 2020

21 - 25
NENA Conference and Expo
Virtual Conference and Expo

October 2020

8 - 8
Webinar: 5 Key Trends in Public Safety & the Technology Implications

November 2020

3 - 5
Critical Communications World (CCW)
Madrid, Spain

5 - 26
Comms Connect Virtual Conference
Virtual Conference and Expo

More Events >

Site Navigation