Delivering Actionable Insights
AIoT is the next generation internet of IoT combined with AI
- By Paul Garms
- Jan 27, 2022
Artificial intelligence (AI) enables machines to learn, understand, and act accordingly – giving it enormous potential. It has developed more in the past five years than in the 50 previous years. From driver assistance systems to smart home appliances and sensors for monitoring for faults in machinery, its presence is felt in many aspects of life.
In the security and safety industry, AI capabilities are just beginning to make an impact but expected to see wider adoption as the industry learns more about the possibilities. The potential includes enabling operators to respond before potential situations occur and assisting managers in identifying and evaluating business opportunities that create new revenue streams or reduce operational costs. This capability requires semantic scene understanding to drive the power to predict events and behavior, something that is achievable with AIoT video systems. AIoT is the next generation Internet of Things (IoT) combined with artificial intelligence (AI). It joins the networking of physical products and the deployment of AI.
AIoT video products add sense and structure to video data. AI capabilities enable the cameras to understand what they are seeing and add meaning to capture video with metadata. This process is an important first step in converting rich contextual and behavioral video data into actionable insights – helping users understand events at an ever-deeper level and predict them in the future. Predictive solutions help users anticipate unforeseen events and prevent them from happening proactively.
Harnessing the power to predict requires making good and efficient use of the rich and versatile data generated by video systems. Video becomes more about intelligent, databased solutions than collecting high-quality images and storing them for the record.
Addressing market challenges with AIoT solutions
AIoT helps strengthen security and safety and enables new uses beyond security. When cameras and AIoT combine, they become intelligent sensors that can provide information on activity or states of objects in an area. These smart sensors enable data-driven solutions that provide business insights to bring new value to organizations.
There are many applications for AIoT video systems, and the possibilities for tailoring solutions to meet specific customer requirements are endless. With machine learning, cameras are trained to recognize objects and situations –for customized solutions that meet the specific needs of users in various markets. Following are a few examples in several markets.
Critical infrastructure. With the latest deep neural network-based video analytics, perimeter detection solutions deliver more accurate person detection capabilities, reducing false alarms. This advanced analytics technology relies on deep learning, which uses artificial neural networks that attempt to mimic the human brain, allowing it to learn from large amounts of data and recognize patterns to tackle more complex tasks faster, easier, and more accurately. This higher level of accuracy reduces false alarms commonly caused by wildlife at the perimeters of remote sites, such as unmanned substations, communications towers, or borders.
Monitoring centers. AI is enabling a transition from human and assisted review of events to a more automated approach. Monitoring centers can benefit from this change through AI-based alarm verification services. These video cloud-based services help to reduce false alarms, increasing the efficiency of operators to ensure that real alarms respond quickly.
Public or commercial buildings. When an AIoT camera detects an object blocking an emergency exit door, it can trigger the public address system to play an automated message over a nearby loudspeaker with instructions to move the object. Alternatively, this same system setup can help users enforce no-parking zones – triggering a message when a vehicle parks or loiters in a fire lane. While these situations are certainly safety concerns, the solutions also reduce the risk of violations and fines when these events occur.
In many larger office buildings, the elevator lobby is the main artery to move people. AIoT cameras can detect crowds in these areas and then trigger other systems to redirect traffic during peak congestion. Making that process faster can have a major impact on the productivity of a workforce and the perception of the accessibility of a building.
Retail stores. AIoT cameras can count the number of people entering and exiting a store to provide operations managers with customer traffic data. This information can help managers understand peak days and times and ensure sufficient staffing to optimize the quality of customer service.
Smart infrastructure. On roadways, AIoT cameras can detect a vehicle traveling in the wrong direction and trigger a roadside unit to notify the wrong-way driver as well as other nearby motorists through dynamic message signs, flashing beacons, or by broadcasting safety messages to smart vehicles. These capabilities provide a real-time safety solution and enable drivers to take action earlier. This same type of system can also alert to slow or stopped vehicles, queues at exit ramps, objects in the road, and other traffic events.
Video sensors can also classify objects as cars, trucks, bicycles, and pedestrians, and detect speed and trajectory – continuously collecting real-time data. Armed with this valuable information, city traffic planning directors and senior traffic engineers can analyze flow patterns on networks of roadways for implementing new policies that result in safer and more efficient intersections.
The latest software founded on deep neural network-based video analytics can even distinguish and classify vehicles in congested scenes. The software can count overlapping vehicles queued at traffic lights or in dense traffic jams, while ignoring common disturbances caused by vehicle headlights, shadows, extreme weather, and sun glare and reflections. With high precision detection, accuracy levels extend beyond 95 percent.
In addition, cities or parking lot managers can take advantage of AIoT video systems to determine how long a vehicle remains in a time-limited space, and alert to vehicles exceeding a certain time threshold to boost revenue streams from parking violations.
This same concept applies at the airport curbside for passenger drop-off and pick up where vehicle-parking time is restricted. AIoT video systems can detect and alert law enforcement to vehicles parked for longer than the maximum time limit.
In lots, cameras can count the number of open parking spaces, or track ingress and egress, and relay this data to the video and parking management systems. Sharing this information and alternative parking locations on a dynamic message sign can help drivers find open parking faster to reduce traffic congestion and emissions.
They can also gather data at airports, where traffic flow and plane status awareness is critical. Data collected could include information on how long an airplane parks at a terminal, the time it takes to load passengers’ luggage, or how long it takes to refuel a plane. This information helps speed the turnaround time for passengers and aircraft, improving efficiencies at the airport, and reducing costs for airlines.
Transforming data into intelligence
Capturing the vast amount of raw data and funneling it into a platform that transforms it into intelligence is essential. This transformation is key to enabling both automated and human decisions. As AI technologies continue to grow, applications are quickly evolving, powering new abilities every day.
AIoT software supports informed decision-making by consolidating and augmenting data from multiple cameras into actionable insights. Users can establish a decision center using software, giving them a single, clear dashboard for evaluation. Dashboards can help users identify unforeseen, unwanted, or future situations faster and more reliably – enabling a response before potential situations occur. They can also deliver business intelligence beyond security.
Capturing data and using AIoT cameras to trigger other systems requires high-quality images to ensure reliability and accuracy. Quality images are essential even in situations where objects are moving, or there are poor lighting conditions or other adverse conditions. If image quality is lacking, the accuracy of the data is at risk, as is any video evidence.
AI technology now in used must also be robust enough to differentiate between genuine events and false triggers such as snow, moving trees, rain, hail, and water reflections that can make video data difficult to interpret. It should also be able to retain information on user-defined objects and situations and refer to these new learnings when processing scenes.
Extending AI to audio
In some cases, video sensors alone may not be enough to provide the needed detection and situational awareness for critical events. Combining audio and video sensors with AI can facilitate a faster response. Audio analytics built in to the camera can recognize and identify the unique audio signatures of sounds like glass breaking, aggressive voices, and gunshots. It can assist operators in detecting, classifying, and locating the direction of the source of an audio event to help security personnel quickly know where and how to respond.
AI drives innovation
As the proliferation of AIoT video systems grows, integrators must anticipate how they will change user preferences. System integrators who understand the full potential and capabilities of AIoT products and software can provide their customers with predictive, sustainable, and trusted solutions that address the challenges users face in their businesses and organizations.
This article originally appeared in the January / February 2022 issue of Security Today.
Paul Garms is the Director of Regional Marketing for Bosch Video Systems in North America.