AI and machine learning are game-changers, and their ability to analyse surveillance video footage is helping people in a multitude of industries to realise a wide variety of goals. This article provides an overview of what intelligent video analytics are and how they work in real-world scenarios.
Automation is the key concept and although human intervention is still required, video analytics are proving to be far more effective than continuous video monitoring by personnel. After all, AI doesn’t easily become distracted and can monitor multiple cameras at once. A study on the efficacy of human-monitored surveillance showed that human observers missed up to 60 percent of targets. That’s a compelling reason to investigate a tech-driven alternative.
In essence, intelligent video analytics analyse what is occurring over time within a specific space. It can recognise and monitor people as well as objects, and it can send out alerts when anomalies are noticed, gather and analyse data, and more.
The best way to illustrate the potential of artificial intelligence video analytics is to examine some of the ways in which it’s already being used.
Intelligent Video Analytics in Practice
Safety, Security, and Quality Control in Manufacturing
As we already noted, intelligent video analytics can recognise and track the movement of individual people and objects. Security is the most obvious application. For example, your surveillance system can trigger alerts when it spots unauthorised people in certain areas of a facility. It can detect objects and determine what is being done with them. And its training tells it what’s normal and what isn’t.
But security is only one facet of what intelligent video analytics can do. It can also help to protect your people from danger. There are often warning signs before a workplace accident occurs, and AI may have a life-saving role to play in spotting risks before accidents happen.
For example, a worker could be in the path of an approaching forklift, or an employee may fail to use important PPE. If properly trained, AI can spot trouble brewing, sounding the alert. When seconds count, early warning can prevent injuries and even save lives.
Besides this, intelligent video analytics outperforms traditional systems used for quality control and defect detection. It can multitask, so it can help with quality control while simultaneously safeguarding workers and assets.
Construction Site Safety
Construction is recognised as the most dangerous industry in the UK today. Monitoring the safety of every activity undertaken on construction sites may not be humanly impossible, but intelligent AI video analytics may be able to change that.
For example, although the laws governing the construction industry include a strong focus on risk identification and mitigation, compliance can be a problem. Besides this, unforeseen risks may occur, and the consequences can be both sudden and deadly.
Although AI video analytics aren’t a panacea, they can certainly help in the quest towards safer construction sites. Once again, anomaly detection and behaviour recognition are at work. These features can spot risks like failure to use PPE and may even identify malfunctioning equipment.
If an accident does occur, video systems can alert the relevant people quickly, speeding response times and potentially saving lives.
Finally, as in manufacturing, smart video analytics will improve site security, identifying unauthorised persons, unusual behaviours, and tracking the movement of materials.
Security Surveillance and Patient Needs in Healthcare
Video surveillance has formed part of the healthcare landscape for decades. Primary concerns include the safety of patients, staff, and visitors as well as the monitoring of scheduled medication to prevent misuse or theft. With intelligent video analytics, these tasks are automated.
Apart from monitoring security, intelligent systems can improve patient care. For example, they can identify unattended patients in distress, or patients who have not been checked on as frequently as they should be.
Additional benefits could include monitoring waiting areas to determine average wait times and any bottlenecks or need for improvement. They can also help in care settings by alerting staff to missed medications, signs of mental distress, or incidents requiring their attention.
Retail Outlets: Customer Behaviour, Sales Optimisation, and Security
As every retailer knows, the way customers behave in store influences what they pay attention to and what they ultimately buy. There’s already been quite a lot of research into this, but intelligent video analytics can uncover important, store-specific details to guide merchandising efforts.
For example, apart from looking at the routes customers typically follow, they can even analyse information such as what they’re looking at, how long products hold their attention, and how frequently signs of interest are followed by a purchase. They can spot times when customers find themselves struggling to find an item too, helping stores to develop strategies that improve customer experiences.
Other valuable functions include monitoring the volume of customers in the store and calculating how long customers generally spend shopping. This helps retailers to ensure that enough personnel are on hand, and helps them to predict peak times and even seasonal changes in customer behaviour.
Of course, security enters the picture, and instead of simply recording what takes place, intelligent video analytics will spot anomalous behaviour that indicates a need for closer monitoring or action.
At the same time, retailers should exercise caution in the degree of surveillance implemented and how it is used. Privacy and human rights concerns are real, and this may limit the degree to which intelligent video analytics can be leveraged.
Traffic Control and Law Enforcement
It’s possible to spot traffic troubles and even prevent them, with the help of intelligent video analytics. For example, when there is increased traffic flow, it can signal a need for adjusted traffic light timing to keep traffic moving smoothly.
When obstacles like stationary vehicles or hazards like dangerous driving are identified, relevant authorities can be alerted so that they can take prompt action.
Intelligent video analytics have a role to play in the design of traffic infrastructure too. Using data on traffic patterns and types of traffic, town and city planners can work on ways to control traffic better, providing a better and safer environment for road users.
Besides this, AI could ultimately do away with the old-fashioned “speed trap” or help authorities to track fugitives from the law. Smart video analytics can identify licence plates, and record traffic infractions, ensuring more effective law enforcement and higher levels of compliance.
Cities can also use video analytics to work on ways to keep pedestrians safer. For example, if jaywalking is common in a certain area, they can determine whether a lack of safe pedestrian crossings is contributing to the problem.
From roads to parking lots and pavements, intelligent video analytics may be among the most effective tools for managing and planning the smart, responsive city that decision-makers have been working toward in recent decades.
Sports, Stadium Management, and Crowd Control
Sports teams are always looking for ways to analyse and improve their performance, and once again intelligent video analytics comes to the fore.
At important matches, crowd control measures are important to the safety of fans, and eagle-eyed cameras coupled with intelligent video analytics can help direct crowd control efforts. Facial recognition can play a role too. For example, if a spectator has been banned from a stadium, facial recognition can spot them at the turnstile, ensuring that undesirable elements don’t ruin the day for other spectators.
Capabilities of Video Analytics Software
Intelligent video analytics software continues to evolve. At present, we can sum up its capabilities by looking at what most systems’ capabilities are. Here are some of the things that artificial intelligence video analytics can already do:
- Detect motion while filtering out irrelevant movements caused by wind, animals, and changes in lighting.
- Identify camera faults or attempts to tamper with a camera.
- Monitor restricted areas for intrusion or entry by unauthorised personnel.
- Identifying moving objects that cross a specified line or boundary.
- Identify movement in an incorrect or unusual direction or lack of expected movement.
- Identify abandoned objects or missing objects that should be present.
- Report on vehicle and traffic volumes and movement patterns.
- Detect unusual movement, e.g. falls, violence, running.
- Generates artificial intelligence video analytics summaries at set intervals.
- Can work with sensors to automatically turn off lights in unoccupied areas.
- Identify trends in building and room occupancy as well as main routes of travel.
These capabilities can be summed up as classification of images, identification of objects (including people), location of objects, detecting objects, and tracking objects. This indicates a wide range of applications across a range of industries. And, with machine learning forming part of the package, video analytics software can be “trained” to function in almost any setting.
How Intelligent Video Analytics Systems Work
Intelligent video analytics processes begin with data gathering – the recording of footage. This makes the choice of cameras and the location of cameras important considerations. The more complete the data or field of vision, the more accurately AI can analyse the data and reach accurate conclusions.
Artificial Intelligence delivers results using its “training.” This begins with existing training data sets with further customisation to suit the context and goals of video surveillance.
The location of the data processing function is also of interest. Central processing indicates that the data is interpreted on monitoring station servers. Edge processing refers to built-in processing functions incorporated into the cameras themselves. Hybrid processing occurs when cameras and servers work together to process and analyse data.
Human intelligence is still an important component in intelligent video analytics. The AI will alert operators to anomalies according to its programming. It will then be up to human beings to decide whether the alert is significant, and whether it indicates a need for action.
Limiting Factors and Challenges
Privacy, Data Protection, and Regulatory Compliance
Compliance with privacy and data protection regulations may pose some challenges. For example, in workplaces, employees must be aware of surveillance and how it is used, and give informed consent. In public places, appropriate signage should alert people to surveillance. In all instances, data must be safeguarded and only be retained for as long as it is necessary to do so.
Facing the Challenges of Complex Contexts
The smart functionality of intelligent video analytics systems requires a large amount of processing capacity. This need increases with the complexity and scale of surveillance activities.
Relevance of Alerts
Intelligent video analytics reports anomalies, but not all of them are relevant or signal a need for action. The software can be trained to reduce the incidence of false alarms, but this must be balanced against signals that may indicate a real need for action.
This may be a sensitive issue, particularly in security-related contexts. For example, a shopper searching for a product may be incorrectly identified as a potential shoplifter. If a response to such an alert is overly robust, reputational harm will ensue.
Potential for Bias
Training data can inadvertently introduce discriminatory bias. For example, if the training data does not include diverse representation, the software may produce alerts based on what it sees as “anomalous” images. Combatting this requires the use of comprehensive training data and follow ups to ensure that bias has not been introduced.
Benefits of Intelligent Video Analytics
While we must recognise the limitations and ethical considerations applicable to intelligent video analytics, it is already being put to work in ways that benefit both organisations and society at large.
Intelligent video analytics can be used to keep people safer, protect organisations from material loss, and reduce the costs of monitoring surveillance footage. It also provides data that can be used to improve human experiences and enhance efficiency. And, when forensic evidence is required, it will be complete and easy to pinpoint despite forming part of many hours of surveillance footage.
If you’re interested in exploring the benefits of intelligent video analytics in your organisation, Advantex, a technology solutions company operating throughout the UK, can help you. Explore the future today. We offer IP Camera and Access Control solutions at the forefront of current technology, helping you to gain its advantages while benefiting from our expert implementation and support. Talk to us about what you hope to achieve and discover effective solutions today.
