- New AI developed capable of recognizing complex visual data
- SMAST can learn and predict complex human actions
- Tool could be used in surveillance, healthcare and autonomous driving, researchers say
Researchers at the University of Virginia School of Engineering and Applied Sciences have taken AI's visual data capabilities a step further with their latest innovation: an AI-powered video analyzer called the Motion-Aware Semantic Spatiotemporal Transformer Network. (SMAST).
This system offers precision in detecting human actions, promising applications in areas such as public safety, motion tracking, and even autonomous vehicle navigation.
At the core of SMAST's capabilities is its ability to process complex video sequences by focusing on the most relevant parts of a scene.
The system integrates a multi-feature selective attention model and a 2D positional coding algorithm with motion recognition. These features work together to ensure that AI can accurately detect and interpret human actions.
The selective attention model allows SMAST to focus on crucial elements, such as a person or a moving object, ignoring irrelevant details. For example, you can distinguish someone who throws a ball from someone who simply raises their arm.
Meanwhile, the motion detection algorithm allows AI to track movements over time, remembering how objects and people have moved within a scene. This gives SMAST the ability to understand the relationships between different actions, making it more effective at recognizing complex behaviors.
In the security and surveillance sectors, the SMAST system can improve public safety by detecting potential threats in real time. For example, it can identify suspicious behavior in a crowded space or recognize if someone is in danger. In the healthcare sector, the technology could be used to track patients' movements, allowing better analysis of movement for rehabilitation or monitoring during surgery.
Researchers say SMAST stands out for its ability to handle raw, chaotic images. SMAST's AI-based approach apparently allows it to learn from data, adapt to various environments, and improve its action detection capabilities. The tool has been put through several academic benchmarks, including AVA, UCF101-24, and EPIC-Kitchens, and performed quite well.
“This AI technology opens doors for real-time action detection in some of the most demanding environments,” said Professor and Chair of the Department of Electrical and Computer Engineering Scott T. Acton. “It's the kind of advance that can help prevent accidents, improve diagnosis and even save lives.”
Through Explore technology