Page 40 - Security Today, October 2019

Smart Infrastructure
data. The result is a distributed network of cameras acting as intel- ligent processing nodes with no single point of failure, delivering a cost-effective and reliable video-as-a-sensor solution. Excellent video images, intelligent event detection and alerts, and data collection and aggregation can all come from the single video device.
With machine learning technology—the latest advancement in video analytics—cameras can be taught to recognize user-defined object classifiers. For example, machine learning can be used to ac- curately count overlapping vehicles queued in front of traffic lights. In many use cases, machine learning capability can improve detection accuracy to enable precise vehicle counts with minimal errors.
Machine learning also enables customized solutions to solve the specific needs of cities and transportation departments. For example, it can detect the formation of icicles or snow buildup on bridges, overpasses and tunnel openings, where they can pose a significant safety risk to motorists or pedestrians passing underneath.
Enhance Safety with Video Analytics
Improve safety by alerting to risks on the road. Intelligent IP cam- eras deliver automatic incident detection and verification for slow or stopped vehicles, queues of vehicles at exit ramps, vehicles traveling the wrong way, objects in the road—such as lost cargo—and other traffic events. Alerts can be sent to traffic management centers or, through integration with highway information solution providers, the IP cameras can trigger third-party systems to notify drivers, im- proving situational awareness.
With video analytics:
• Early incident detection enables traffic and transportation op- erations centers to implement the necessary workflows to resolve road irregularities faster, enable first responders to intervene more quickly, and avoid secondary accidents.
• Integration with dynamic message signs, and dedicated short- range communication (DSRC) broadcast messages to smart ve- hicles, ensure drivers are instantly alerted to safety issues.
• Detection of pedestrians in a crosswalk can alert the traffic con- troller to preempt traffic signals to increase safety, enabling smart intersection control.
• Jaywalkers at night can trigger the IP camera to activate an on- board illuminator to make pedestrians more visible to motorists.
• Integration with highway information solutions can broadcast mes- sages to smart vehicles to alert them to the presence of pedestrians in the area the vehicle is travelling and display the pedestrian’s GPS coordinates on a live map in the vehicle’s on board unit.
Analyze Data to Extend Beyond Safety
With video analytics, the IP camera becomes an intelligent sensor that can classify objects as cars, trucks, bicycles, and pedestrians, and detect speed and trajectory. Object classification enables the cameras to recognize what they are viewing for data-gathering purposes.
Using video as a sensor and software tools for tapping into the camera generated metadata, the data can be extracted and stored in relational databases allowing city traffic planning directors and senior traffic engineers to continuously collect real-time data to analyze flow patterns on networks of roadways. The camera-generated data can be used for implementing new policies that result in safer and more efficient intersections. This data can also help them determine how
pedestrians use certain locations to provide insights into possible safety improvements.
Multiple cameras can feed data to dashboards that deliver action- able information for business intelligence purposes—helping city planners understand traffic patterns, congestion points, and more. Data examples include:
• Pedestrian, bicycle, and vehicle counts
• Classification of vehicles, such as cars versus trucks
• Average speed and direction
• Road occupancy
Infrastructure owners can choose whether they want both video and the data or just the data. When only data is needed, low band- width connections can stream it from the sensor into the data ware- house, reducing network impact and allowing video to stay private.
In busy cities, video analytics can also help monitor parking lot occupancy and curbside parking as well as help enforce no-parking zones. In lots, cameras can count the number of open parking spaces, specialized spots—such as those for handicap or electric vehicles—or track ingress and egress. The cameras then relay this data to the video and parking management systems. Sharing this information, along with alternative parking locations on a dynamic message sign, can help drivers find open parking faster, getting them off the road and reducing traffic congestion and emissions.
For smart parking applications, machine learning adds the advan- tage of determining the amount of time a vehicle has been parked and much more. For example, at the airport curbside for passenger drop-off and pickup, vehicle parking time limit is restricted; machine learning can detect and alert law enforcement to vehicles that exceed the maximum time limit. Machine learning analytics technology does not have a timeout dependency as compared to standard analyt- ics detections of parked vehicles.
Proven Solution
While these types of capabilities and solutions may seem to be part of futuristic solutions, smart connected city and transportation so- lutions are already here and are being used in states and provinces throughout North America, whether through full implementation or in pilot projects.
For example, in one Midwestern state, IP cameras with built-in video analytics are helping to improve highway safety by warning drivers of dangers ahead. Some of these alerts include cross-traffic warnings, curve-speed warnings, pedestrian detection, queue warn- ings, work-zone warnings and wrong-way driver detection. Events detected by the cameras trigger the highway information solution to send messages about these incidents directly into the display units in connected smart vehicles. For the majority of unconnected vehicles, messages are broadcast to dynamic signs and other alert beacons.
From improving traffic flow on highways to implementing proj- ects for pedestrian safety initiatives and gathering data for smart in- frastructure planning, video sensor technology enables a wide range of customized solutions to meet the needs of city planners and senior traffic engineers around the world.
Joel White is the North America vertical marketing manager for Intelligent Transportation Systems and Smart Cities at Bosch Security and Safety Systems.
GS14
GOVERNMENT SECURITY OCTOBER 2019