Topics: Coronavirus | AI | Data Science | Deep Learning | Machine Learning | Python | R | Statistics

KDnuggets Home » News » 2020 » Sep » News, Education » NIST $240K Challenge: Saving Lives, One Pixel at a Time ( 20:n34 )

NIST $240K Challenge: Saving Lives, One Pixel at a Time


Video analytics that could save lives and property are just out of reach. A new prize challenge, Enhancing Computer Vision for Public Safety, is designed to help develop a new line of research that will bring such tools closer to reality.



Sponsored Post.
NIST Dataset Challenge

 

Video analytics that could save lives and property are just out of reach. A new prize challenge, Enhancing Computer Vision for Public Safety, is designed to help develop a new line of research that will bring such tools closer to reality.

First responders operate in environments that challenge cameras with rain, dust, smoke, low light, and unstable camera mounts. Image quality plummets, and video analytics become less reliable.

The solution is a metric that detects image and video quality impairments and enables a computer vision system to respond with diverse strategies—adjusting camera settings, applying a noise reduction filter, marking regions where computer vision is likely to fail, or switching between computer vision algorithms.

The goal of the Enhancing Computer Vision for Public Safety Challenge, hosted by the Public Safety Communications Research division of the National Institute of Standards and Technology and challenge partner First Responder Network Authority, is to inspire new research into the relationship between camera impairments and computer vision.

By making these computer vision systems smarter, emergency operations can become safer.

Beginning Sept. 8, NIST invites contestants to submit a concept paper proposing a method to measure the failure rate of a computer vision algorithm. This is the core dilemma: how to quantify the impact of camera impairments on computer vision. Contestants must also propose a small dataset of images or short videos that will be used to demonstrate the failure rate measurement method. These media must contain actual camera impairments, not simulations. Concept papers are due by Oct. 20, 2020.

Winning concept papers will be invited to compete in Phase 2 where contestant teams will implement their proposed solutions. Each contestant (teams or individuals) who enters will compete to win up to $28,000 each and help contribute to further computer vision research.

Learn more about the problem statement, the challenge requirements, and total prize purse at the challenge website. Get ready for the challenge now, and plan to submit your concept paper to psprizes@nist.gov starting September 8th.

Image

Sign Up

By subscribing you accept KDnuggets Privacy Policy