HOUSTON--(BUSINESS WIRE) -- Behavioral Recognition Systems, Inc. (BRS Labs) announces the release of the latest version of AISight, the first and only Behavioral Analytics solution. This latest release, AISight 2.1, addresses two of the industry’s most difficult problems: accurately classifying objects such as humans, animals, vehicles etc., and efficiently recognizing and tracking large numbers of moving objects in the camera’s view.
There are literally dozens of patents filed explaining how a camera should recognize the human form. The simple fact is that a human will look different from one camera to the next. For example, if a camera is mounted 20 feet off the ground, it might not even see the legs of a person walking by. To identify objects accurately, the vendor must adjust and program for every possible camera angle - a time consuming and expensive task which has never been adequately executed.
Artificial neural networks simulate the functions and processes of biological neural networks. With its new micro-classification feature, AISight now classifies objects without any pre-programmed definitions or specifications, replicating the neural processes of a human. AISight classifies every object in the camera’s field of view based on observations and memories the Machine Learning Engine (the brains of AISight) creates autonomously. AISight classifies and determines the differences between humans, cars, animals and all other objects. The system categorizes and classifies observed objects according to each camera’s unique field of view.
In other words; AISight establishes its own criteria for classifying a human (or any object), specifically tailoring it to each camera’s field of view. In addition to observing and describing behaviors, the expanded learning engine of AISight 2.1 leverages refined micro-classification features to gather information about dominant object content, including a subject’s size, color, reflectivity, sheen, shape, and whether it occurs singly or as part of a group. Tracking separate objects and groups of objects in real time, the micro-classification feature enables the system to observe the scene and learn to identify not only normal and anomalous behaviors, but also to identify the types of objects that exhibit those behaviors in the scene. This provides AISight with a wider and broader understanding of scene content to yield highly effective identification, analysis and tracking of behaviors.
"Analysis based on micro-classification provides substantial benefits to security environments. Not only does AISight have the capability to observe a human in a restricted area; AISight is also able to identify other elements such as hue, color and saturation of the clothing the human is wearing. This provides critical data that can be used to determine if the human was clothed in a uniform common to staff allowed inside that area. This is particularly useful for applications where access is restricted, but certain types of human or vehicular traffic is frequent," said Ray Davis, CEO of BRS Labs. "This new descriptive or adjective-based identification architecture also makes AISight 2.1 highly effective in monitoring the activities of groups. Frequently, the first sign of an event unfolding is the scattering or converging of a crowd. AISight can learn patterns of activities for groups, including sudden or abnormal changes in movement as the group travels through the scene, alerting on abnormal activity in real time to security command and control systems. Only by applying Artificial Neural Network (ANN) Technology can a system have the power and ability to perform these functions accurately. After over 150 Man-Years of development, BRS Labs is proud to provide the industry with a truly intelligent system to fight crime and terrorism at home and abroad,"
Current development includes integrating the cognitive machine learning core with a variety of other perimeter intrusion detection sensors; allowing AISight to analyze data it receives from sources beyond simple video, such as Forward Looking Infrared (FLIR) camera sensors.