CoVar is a leader in the nascent field of computer vision.
Our technologies leverage widely available video feeds from commercial cameras to extract useful and actionable intelligence in real-time. Video data is a very information-rich, yet highly unstructured source for sensing data. CoVar’s computer vision software automates human-like information extraction from video, so a computer can reliably organize and extract information found in video feeds. Our applications for computer vision include monitoring and tracking personnel in industrial settings, closed loop machinery monitoring and control, and quality control and measurement for materials inspection in the commodities field—to name just a few. With our leading capabilities, home-grown proprietary techniques, and extensive real-time implementation architectures we are providing innovative new capabilities to our customers, saving time, money, and lives.
Computer Vision for Safe and Efficient Drilling
Automated interpretation of video data is known as computer vision, and recent advances in computer vision technologies have led to significantly improved performance across a wide range of video-based sensing tasks. CoVar has developed numerous computer vision applications for improving the process of drilling for oil and gas. Modern drilling involves scores of people and multiple inter-connecting activities. Obtaining real-time information about ongoing operations is of paramount importance for safe, efficient drilling. By applying automated, computer-based video interpretation, continuous, robust, and accurate assessment of many different phenomena can be achieved through pre-existing video cameras. CoVar is developing the next generation of computer vision techniques to improve safety, reduce costs and improve efficiency.
Detection of Roadside Hazards through Visual Change Detection
Computer vision can be used to aid soldiers in finding emplaced explosives when leading convoys through enemy territory. When the same route is traveled repeatedly, video from prior traverses can be compared to the current video feed and changes that might indicate that an enemy has emplaced a roadside bomb can be automatically flagged. The software must be smart enough to compensate for varying lighting conditions, shadows, camera orientation and exact placement, etc. Our software automatically identifies significant differences that are consistent with an emplaced explosive, alerting the soldier and highlighting the object on the screen. CoVar employs a combination of temporal- and spatial-video warping and feature-point-based change detection models using variational Bayesian statistical models to achieve robust change detection in widely varying scenes.