Identify suspicious behavior in real-time. Get fast, accurate pre-threat alerts.
Simple, scalable, automated surveillance – every camera, every minute, every day.
BRS Labs' AISight® system is an artificial intelligence solution for video surveillance that teaches itself to recognize and alert on suspicious behavior across any camera's field of view. AISight installs and is productive within days – without custom programming or human intervention. This highly automated solution works with a wide range of cameras and networks, in new or existing infrastructures, to dramatically improve situational awareness and risk assessment for organizations seeking real-time threat recognition and response.
Scalable – AISight scales easily to accommodate multiple locations and facilities. It handles massive amounts of live and recorded video data with exceptional accuracy and without complexity.Customers gain the ability to recognize and respond to suspicious or inappropriate behavior in real-time on every camera in their network, up to thousands of cameras in total.
Intelligent – This scalability comes with a much simpler deployment scenario than other technologies, since they don't have artificial intelligence and behavior recognition at their core. AISight automatically teaches itself what's normal within every camera's individual field of view. There's no need for rules, masks or custom programming. Simply plug in a new camera or move an existing one, and the system updates itself.
Fast – AISight also simplifies the processing of huge volumes of video, automatically alerting in real-time for live feeds or later for recorded footage. Users get direct visual replay and feedback alongside real-time alerts for suspicious behavior, so that they can make fast, accurate decisions on the severity of a threat and determine the most appropriate response.
Accurate – Since AISight automatically builds bodies of evidence for its alerts, there are far fewer false alarms than with other technologies. This ability to recognize suspicious behavior helps stop threats before they become crises, which in turn saves money, prevents injury and protects critical facilities from theft and abuse. Users operate much more efficiently, since they only have to look at the video they truly need to see, exactly when they need to see it.
Robust – Built-in redundancy capabilities ensure that all alerts and relevant video footage are retained, even during service interruptions between a camera and the AISight server. No data is lost, no matter how large the deployment.
There is a better way to enhance safety and security, improve real-time response to unfolding events and do so while saving both time and money – AISight, from BRS Labs.
Installation
Whether you are installing AISight on a network of 100 or 10,000 cameras, the installation is the same; fast and easy! Where other systems require pre-programmed rules, trip lines, zones, masks and other expensive and time consuming setup for each camera; AISight is installed in just a few hours because it learns on its own what should and shouldn’t occur in each camera’s view.
- View More
Without human intervention, pre-programming or setup time, AISight provides:
- Unsupervised learning based classification – AISight uses self-organizing memory maps to recognize common patterns among the features of observed subjects and defines types based on those patterns. This allows AISight to differentiate between behaviors of objects such as humans and automobiles, and understand their relationships
- Classification anomalies – AISight recognizes when subject types match no known pattern of appearance/properties and issues an alert to that effect. This will effectively notify you if a foreign object is introduced into your environment
- Position-based activities and motion – By learning to recognize where it is common for things to appear, disappear, enter, exit, stop, start, turn, move quickly/slowly, accelerate quickly/slowly, AISight alerts on position-based activities and motion
- Spatial anomalies – AISight uses spatial memories (position-based maps) to recognize anomalies for behaviors that violate the learned patterns regardless of where they occur in the scene
- Time-of-Day anomalies – AISight uses temporal memories to issue alerts on anomalies for behaviors that violate the learned patterns based on time-based maps. AISight learns to recognize where and when it is common for things to appear, disappear, enter, exit, stop, start, turn, move quickly/slowly, accelerate quickly/slowly, and reports on activities that are not normal given the current time of day even though it is normal for that activity to occur at a different time.
- Day-of-Week anomalies – AISight uses spatial memories to issue alerts on anomalies for behaviors that violate the learned patterns in its Time-based Memory Maps. This allows for normal activity that occurs on one day of the week, but never on another, to be tracked and reported
- Autonomous maturity recognition – AISight determines on its own when it has achieved sufficient scene knowledge and then begins alerts by itself
- Multiplatform Desktop Applications – AISight can run on 32 & 64 bit Windows & Linux
Maintenance
By removing the need for individual camera setup, AISight has also reduced the cost and resources needed to maintain an intelligent video system. Where legacy systems would require an engineer to reset trip lines, zones, or masks every time a camera was moved or the environment it was viewing changed (by humans or weather); AISight constantly monitors each scene and automatically adjusts and learns new or changing environments.
- View More
To insure that modifications, adjustments or other engineering expense is not necessary, AISight accomplishes the following without human intervention:
- Sudden illumination change – AISight detects sudden changes in illumination and reports if it is at an inappropriate time or place; otherwise it compensates to continue monitoring
- Image stabilization – AISight compensates for image distortions, jitter and affine transformation distortions in the video stream to better manage cameras subject to vibrations or slight movements
- Luminance filter – AISight compensates for illumination changes over regions, reducing the effect of shadows, and enhancing contrast
- Pixel noise filtering – AISight uses pixel stabilization to reduce the effects of noise in the received video frames
- Dynamic color balance – AISight can be configured automatically to apply color and gain correction to video frames
- Video QOS reporting – AISight reports information on quality of service for the video stream
- Optional resolution sizing – AISight can be configured automatically to reduce large format video down to manageable sizes. AISight normally operates on SIF/CIF resolution.
- Optional frame-rate throttling – AISight can be configured to automatically reduce large frame rates down to manageable rates. AISight normally operates at 5-10 Frames per second
- User Groups and User Assigned Security Roles and Role-Based Permissions – Access to feeds can be restricted to selected users or user groups
- PTZ camera support – AISight supports pan-tilt-zoom cameras in tour-of-duty mode and the use of presets by building unique memories for each location the camera sees
- Save and restore of behavior memories – AISight will automatically save its state frequently to protect the information that it has learned. At system startup, existing memories are restored.
- Activity governed decay and forgetting of memories – AISight employs activity governed decay for the forgetting of memories to auto-forget behaviors that are no longer common – precluding the possibility of over learning
- Learning visualization – AISight provides dozens of images and reports to allow users to understand the nature of the information that the system has learned. This gives management the information they need and operations the assurance that the system is if full operation.
- Camera/feed performance monitoring and control via System Manager – AISight’s web based client allows users direct control of monitoring and alert publication/dispatch criteria
- Video Feed Groups - AISight has the capability to perform most feed related operations on the entire group of cameras
Accurate Relevant Alerts
The objective and value of any video surveillance system is to provide situational awareness to either prevent or respond to terrorism, crime or other potentially dangerous activity. Analytics utilize trip wires and other tactics to try and catch the bad guys, however these methods have been known to produce consistent false alarms due to shadows, glares, leaves blowing in the wind or other events of this nature. AISight generates alerts when something is unusual … It’s as simple as that. There are no false alarms and the volume is always manageable because the system constantly monitors itself as closely as it does your cameras.
- View More
- Alert archiving – AISight has an archiving and viewing tool for allowing user to generate custom archives of alerts meeting selected criteria
- Live monitoring of tracking and alerting via operator console – AISight’s desktop application allows users to view live tracking from camera feeds, monitor alerting, and start/stop feeds
- Time-based control of alerting – AISight gives users the ability to define precisely what types of alerts they would like to be informed of and at what times of the day
- Alert Criteria – AISight allows users to apply fine control to what behaviors will be captured and reported-on by the system
- Alert Annotation – AISight can be configured to allow users to add comments to alerts
- Control of Alert Volumes – AISight provides separate control of alert volumes for publishing (forensic) and dispatching (email) purposes.
- Printable Alert Details – AISight has the ability to print full alert metadata including image for environments where network connectivity is limited
- Configurable Alert Notification – AISight notifies the user of alert arrival by sound and/or popup (Operator Console)
- Position-based appearance-features – AISight learns to recognize where it is common for subjects to have certain common features – colors, shapes, movement patterns, shininess, structures, etc. This allows AISight to understand what types of objects are allowed in every area of the camera’s field of view and what objects aren’t
- Trajectory – Every aspect of an objects trajectory is analyzed and memorized – Using data from the spatial activity and motion memories, from semantic representation of motion via phase space partitioning maps and ART networks
- Alert volume normalization – AISight continually re-evaluates the statistical significance of the various anomalies it observes and adjusts its alerting criteria accordingly
- Loitering – AISight is able to recognize those locations within a scene where loitering is unusual and will alert if it is observed
- Discarded object anomaly – AISight is able to recognize stationary, discarded, or abandoned objects identified in unexpected locations within a scene
- Tailgating anomaly – AISight is able to recognize unusual pairs of nearly matching actions, for example “tailgating” through entrances
- Long-term memories – AISight maintains long term associative memory for events that have occurred repeatedly and have been reinforced. This allows common activities to be recognized as common.
- Episodic Memory – AISight uses midterm memory for events occurring in the fairly recent past just like the human brain
- Perceptual Associative Memory – AISight utilizes short term memory to track and report events that have just occurred
- System anomaly diagnostics – AISight reports on camera outages, loss of video signal, obscured cameras, poor signal quality, unstable scene, field-of-view-change, change from preferred preset, etc.
- Field-of-view change – AISight recognizes when a camera is moved to point at something else and alerts regarding this
- Camera occlusion – AISight recognizes when the camera has been obscured and alerts regarding this
- Camera-out-of-focus – AISight recognizes when a camera is out-of-focus and alerts regarding this
- Alert directives – AISight provides a mechanism for end-users to override decision of AISight and insist that certain behaviors are Always or Never alerted on
- Focused alert directives – FADs are an enhancement of alert directives that allows for more control of the spatial locations to which alert directives will be applied
Benefits
- Enhances situational awareness and risk assessment by applying behavioral recognition technology that's easier to deploy, easier to use, more effective and less expensive to operate than any alternative currently available
- Teaches itself to recognize suspicious behavior within any camera's field of view without custom programming or human intervention
- Uses both new and existing cameras and networks, which dramatically reduces the time and expense associated with deploying and operating intelligent video surveillance
- Nearly eliminates false alarms, making video surveillance a true real-time solution for intervening as events unfold, rather than a forensic tool used most often after the fact
- Delivers direct visibility into safety issues as well as security, such as automatic recognition of smoke, fire, flooding, static objects vibrating or in motion, etc.
- Integrates easily and transparently with a wide range of VMS and PSIM solutions, including Cisco VSOM/VSMS, VidSys VidShield and RiskShield, Milestone XProtect, Lenel OnGuard, ONSSI Ocularis, Genetech Omnicast, Geutebrück GeViScope and many others
