Research

Your are currently viewing: Home

Research

The following is a list of active or recent projects. There is also a list of previous projects.

Posture Recognition for Automated Evaluation of MOUT Training

Detecting and classifying marines and their postures is critical for automated training performance evaluation. The main objective for this project is to detect articulated objects in various configurations in still images, with an unconstrained, cluttered background that can not be modeled. Our approach usesa multi-class boosting procedure that favors discriminative features shared by multiple objects and views. Unlike traditional methods, this approach scales logarithmically with the number of classes. Click here to see a video.

Improving Crew Threat Awareness through Augmented Vision

The focus of this research and prototype system development is to integrate spatially related data into a synthetic view of the outside environment for use by vehicle commanders, via a visually augmented indirect vision system. Street names, building information and intelligence data will be fused with the video from vehicle-mounted cameras. Terrain-associated knowledge hence persists in place in the environment, rather than being verbally relayed, stored in text documents or on paper maps, or being lost entirely. Crucial information - unobtrusively displayed at the right moment and place - allows a vehicle crew to better understand their operational environment, to be aware of threats that may be present, and ultimately to improve situational awareness and crew safety.

AR-VAST

The Augmented-Reality Virtual At Sea Trainer (AR-VAST) allows SWO training for difficult situations such as small craft swarm attacks. The goal of AR-VAST is to provide the Navy and Marine Corps with training capabilities to defend against small boat attacks and for operations in littoral environments.

“Training on crew served weapons on coalition ships has not advanced to meet this new threat,” according to Lt. Nate Conger’s thesis abstract regarding AR-VAST. “The creation of a prototype Augmented Reality Virtual At Sea Train shows that emerging technologies in augmented reality systems can be used to create a low cost effective answer to this training need.”

A Common Body Gesture Description Language

There are many ways to describe human body postures and gestures in the various related fields of human-computer interaction, psychology, computer graphics and animation, motion capture, and so forth. This research investigates how these different fields can directly benefit from each other's advances by standardizing how vocabulary is introduced into a language that can serve all related disciplines.

Embedded Computer Vision

Performing computer vision on PCs is extremely power hungry. Embedded vision systems achieve greater efficiency and promise mobility, possibly even higher performance. We are currently exploring these opportunities for various applications, including robot vision, networked camera systems, and vision for handheld devices.

HandVu: Vision-based Hand Gesture Recognition and User Interface

With almost any color camera and sufficient processing power, this software collection implements a hand gesture interface. HandVu detects the hand in a standard posture, then tracks it and recognizes key postures - all in real-time and without the need for camera or user calibration. The output is accessible through a client-server infrastructure in a custom format and as OSC packets. Read more about HandVu or even download the alpha-stage software.