MU researchers develop object recognition software
The research can detect if something is a person or an object.
Published Nov. 3, 2009
MU researchers are working on developing software that can identify humans and objects within videos.
The software would enable computers to search inside videos and detect humans and specific objects and perform other video analysis tasks, an MU News Bureau news release stated.
"The goal of our research is to improve how computers interpret the content of a video and how to identify it," said Tony Han, electrical and computer engineering professor in the College of Engineering, in the release. "There are lots of possibilities with video-based detection, and it could come at quite a low cost compared to object and human detection using other sensors, such as thermal sensors."
Intelligent video surveillance requires human and object detection, the news release stated. If a security camera captures an image of an injured person lying on the ground, the computer would not only store the surveillance image, but also send signals for help.
Human detection software can be used in many areas the news release stated. It can be applied to assisted driving and the care of older adults living at home.
"My students and I are working on algorithms for automatic object detection, but these are very difficult to perfect," Han said in the news release. "We're trying to find a way to create reliable detection algorithms, but it takes a lot of time to test them. We have manually labeled more than 3,000 images with object locations and have used them to test our algorithms."
According to the release, Han and his students attended a challenge in object detection this fall, where they competed in detection for objects in 20 categories against researchers from all over the world. In their first time competing, they won first place in detection of potted plants and chairs and second place in detection for humans, cars, horses and bikes.
Kelsey Jackson, a senior information specialist at the MU News Bureau, said a small group of students and Han are working on projects related to object detection.
"For each sort of object, they do an algorithm," Jackson said. "So some students may be working on people while others are working on objects."
Jackson was not sure when a patent for the technology could be expected but said the research had been published in numerous publications.