A University of Houston computer science professor and the City of Houston are collaborating to increase the awareness of security cameras by alerting an operator if there are mechanical issues with the camera or if it is recording unusual activity in its view.
Professor Shishir Shah and his research team received a $1.57 million grant from the National Institute of Standard and Public Safety Innovation Accelerator Program to test algorithms that can alert first responders to potential emergencies, according to a UH press release. The goal is to alert help if it’s needed at the camera’s location.
“Can you design algorithms that say, ‘Look at this camera that went out of focus, tell someone maybe that needs to be fixed’?” Shah said. “ So that’s part of what we are trying to look at as part of this research project, and hopefully design some useful algorithms that can automatically provide alerts to people for scenarios where cameras are not functioning properly.”
Shah’s research is a part of the computer vision field, where inferences on what object is seen by a camera sensor are made.
“So obviously our brain has figured out how to make those associations. The questions is, can we understand how that happens so that maybe we can have that same process replicated on computers,” Shah said.
The City of Houston will be working with Shah and his research team, allowing them to use their near 900 cameras throughout the city, to one day adopt the algorithms if proven useful.
“Our goal is to see if they can come up with an analytics solution that is viable, that we would end up adopting and using permanently to assist in better and more quickly identifying a situation that’s occurring,” said director of the Public Safety Video Initiative Program Julie Stroup, who will be working with Shah.
Using the City of Houston’s security cameras, rather than just the ones at the University of Houston, will allow the researchers to test more situations than those that occur just on campus, Stroup said.
“We are going to be working and collaborating closely with the City of Houston,” Shah said. “We will basically be talking to people there to understand where we can test our algorithms, how can we go about testing our algorithms, and with their help, try to understand what the impact might be.”
One issue this research may eliminate is the daunting task of monitoring hundreds of security cameras, which have uneventful feeds for most of the time, Shah said.
“It’s physically impossible for people to view all of those cameras at the same time,” Stroup said. “What this would allow us to do is pop up things in front of people and say, ‘Hey, we aren’t sure, but there might be something here you want to look at,’ and allow an operator to look at it and see if there is some type of an event occurring.”