Video Artificial Intelligence
The IntelliView Vision System (IVS) — an IIoT analytic imaging and alerting platform — is the product of over A MILLION hours of rigorous field testing, customer use, and ongoing refinement. The IVS utilizes the combined capabilities of patented and proprietary software, supervised machine learning, deep neural networks, distributed processing and the latest sensors.
The convergence of these technologies makes it possible to deliver enhanced detection accuracy and high analytic performance, with a low incidence of false-positives. IVS is able to solve the unique, mission critical problems of today’s industries, such as above-ground liquid leak detection, remote equipment inspection, and perimeter surveillance for oil & gas and mining companies.
Analytic Software Capabilities
> Live Bi-Spectral Sensor Input Analysis
The IntelliView analytic software is applied in real time to the thermal and full HD color video feeds of the DCAM™ alert data, live feeds, videos, tools, features, and settings — doubling the system’s detection capability. The Long Wave IR sensor (uncooled microbolometer) tracks emitted energy without the requirement for lighting, and allows the analytics to perform well even in adverse weather.
> Dual-Sensor Detection Correlation
An event is analyzed and matched by the DCAM’s thermal sensor and its color sensor, independently. This provides an additional layer of event qualification at the network edge, which helps to reduce false alerts.
> Multi-Data Processing
Various types of data (eg. radiometric, environmental, user input) are analyzed and presented at the System Console as information that can be acted upon by the monitoring personnel and first responders.
> Multi-Region Detection
Independent analytic rules can be implemented for different areas of a camera view, providing comprehensive coverage for complex and unconventional sites.
> Analytic Rule Coupling
Two analytic conditions from either or both sensors of a DCAM™ can be configured to work in tandem, improving object detection and adding a layer of validation.
> AnalyticControl of Digital I/O Devices
Peripheral devices can be remotely activated and deactivated by analytic rules.
> Object Characteristics Specification
The analytics can detect objects of interest based on their unique properties (eg. size, temperature, speed, color), disqualifying objects that fall outside of those parameters.
> Object Validation by Classification
Deep learning artificial intelligence implemented on incoming alerts provides an extra layer of object qualification on the basis of its type (eg. people, car, animal).
> Low False Alert Levels
The combination of hardware and software technologies deployed in an IntelliView system enables the analytics to perform optimally and accurately in a wide range of conditions, including extreme temperature (hot and cold climates), indoors, outdoors, and hazardous environments.
> Environmental Filtering
Analytics mitigate the impacts of weather and ambient elements (eg. glare, shadowing, heavy rain, snow, and fog) that are common causes of false positives.
> Image Stabilization (IntelliView proprietary software)
With proprietary software from IntelliView, image distortion from camera shake — typically caused by strong winds — is automatically corrected to help maintain optimal analytic function.
> Detection Sensitivity Control
Detection levels can be adjusted to suit a specific environment and to meet industry accepted false positive rates.
Implementation of IntelliView Analytic Software Technologies within an Industrial IoT Architecture
Onsite (Network Edge)
IntelliView’s image processing technologies, which are built into the DCAM™, process thermal (LWIR) and color video feeds in real time. The system memorizes the background and adjusts to background changes. When an event that meets user-specified conditions is detected, this is reported with image and video using minimal network bandwidth to the System Console located at the customer control or monitoring center.
At The Monitoring/ Control Centre
The System Console (SC) has the ability to further evaluate an alarm event using supervised machine learning, employing deep neural nets akin to the systems used in self-driving cars. The incorporation of adaptive artificial intelligence trained on pre-classified data adds a layer of improved recognition and processing of detected events to the analytics suite. System alerts, data and settings are also accessible via a secure web app.