Nick Herbert is the Global Head of Government & Public Safety for Fujitsu, enabling a trusted digital society.
getty
I spend much of my working life focused on improving public safety and security for governments, organizations and individuals. As expected, the interest and investment in technology to improve society’s safety is increasing daily. Video behavior analytics (VBA), powered by artificial intelligence (AI), is a leading tool in this transformation. It has such potential for good that I believe we all have a responsibility to share our knowledge to strive towards a better, safer society for all.
VBA works by detecting and interpreting human behavior and intent from video streams, transforming safety measures across various industries, from transportation and retail to government facilities, but many are not yet on this journey.
Understanding Video Behavior Analytics
Video behavior analytics involves the use of AI algorithms to analyze video footage and identify humans and their actions to recognize patterns that may indicate suspicious or dangerous behavior. Unlike traditional video surveillance, which relies heavily on human monitoring, VBA automates the detection process, allowing for real-time alerts and quicker responses to potential threats, ideally preventing security incidents before harm is caused. This technology can detect a wide range of behaviors, from aggressive actions and vandalism to individuals entering prohibited areas or showing signs of distress requiring medical assistance.
In 2021, 48,830 people in the U.S. died of gun-related injuries, and 22 million Europeans were physically attacked. Thousands of people are hurt on train lines per year, with 8,000 people in India in 2020 as a key example. Could early video detection of activity reduce this number?
The Impact On Public Safety
The implementation of VBA in public safety is multifaceted, addressing several key areas:
Transportation
One key area is increasing the safety of passengers and drivers by identifying aggressive behavior, vandalism or unauthorized access to restricted areas, such as train tracks. By detecting these behaviors early, transport control centers and station staff can take preventive measures or respond to incidents, minimizing risks and enhancing overall safety.
Government Facilities
Ensuring the safety of citizens within government facilities is critical. VBA can detect threats of suspicious behavior or unauthorized access to sensitive areas, allowing security personnel to step in before incidents escalate.
Educational Institutions
Violence in schools has been well publicized, especially in the United States. Campus safety could be improved with VBA by identifying potential threats, such as unauthorized intrusions, aggression or potential bullying. This use case helps to promote a safer learning environment for students and faculty staff.
Public Events
Large gatherings in public spaces or arenas, such as concerts, sports events and festivals, create unique security challenges. VBA could help monitor crowds for signs of aggression and identify individuals entering restricted areas, left luggage or potential overcrowding.
The Technical Challenge
Many real-world applications of detecting human behavior encounter environmental conditions, which necessitate a flexible approach. A "simple" single-model approach often struggles with precision and recall, especially when dealing with occlusion, where an object or person is partially obscured. Additional issues, such as poor lighting or differences in behavior between "attendees" and "staff in uniforms," call for a more nuanced solution.
By integrating multiple AI models into a multi-modal approach, it is possible to significantly enhance results. For instance, large language models (LLMs) designed for image and video recognition can complement skeletal models used for AI human behavior detection, which can be complemented again by tracking models that can "follow" a person through multiple camera feeds.
The combined approach not only helps improve the overall accuracy but also ensures more robust handling of challenging scenarios, such as poor lighting conditions or crowded environments. Putting tech aside, the aim here is to have a solution that delivers high precision and recall. This ensures that businesses and public sector organizations can deploy more effective systems and trust that the alerts generated indicate situations where action is necessary to enhance safety and security measures.
The Broader Implications For Society
The initial focus on building a business case for vision AI in public safety is often the reduction of operational costs. However, for me, the biggest benefit is in the way that this technology can impact the individual—staff, customers and citizens. By reducing the burden on frontline workers and improving response times, VBA can contribute to being a cornerstone in our public safety infrastructure fabric.
Automating routine tasks can help to reduce stress and mental health issues among employees, allowing them to focus on critical events and alerts. In sectors facing labor shortages, like retail and transportation, VBA boosts workforce productivity, maintaining safety and security and alleviating the burden on frontline workers. Additionally, optimizing monitoring processes can reduce costs by up to 30%, making it a vital component of public safety infrastructure.
Video behavior analytics lays the foundations for a significant step forward in how we improve public safety as a society. Yet, there are privacy concerns and ethical considerations for implementing any computer vision solution within a public domain.
These concerns relate to the installation of video behavior analytics in open public areas, which will trigger privacy and ethical questions. Surveillance inherently implies that people are being watched continually, and though it may be legal, it can be invasive to an individual and without explicit consent. Additionally, there is the threat that such large volumes of data collected could be exploited, whether by external actors or by repurposing the data into other unrelated initiatives. The other often-cited issue is bias. Any AI system has the potential to lead toward bias, only exacerbating existing prejudices and providing unfair outcomes for some groups or communities.
To overcome these obstacles, organizations must adopt a proactive and transparent approach. Explainable AI, with transparent policies and rules, is essential to build trust. It is also important not to forget the value of the "human in the loop." This means that even though AI is used to recommend actions, the final decisions belong to humans—especially when it comes to public safety, security or other sensitive arenas.
For organizations seeking to implement this type of solution, the importance of outside opinion, stakeholder engagement and community outreach will be a key pillar of your adoption program. No system is perfect, and compromises are often needed. What better way to agree on the priority areas than with the citizens that you seek to serve?
In Summary
By creating an open dialogue and sharing approaches to challenges and successes, we can collaborate between the public and private sectors to advance our application and understanding of the ethical applications of this technology to harness the potential and create a safer, more secure world for all.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
{Categories} _Category: Takes{/Categories}
{URL}https://www.forbes.com/councils/forbestechcouncil/2024/09/03/the-future-of-public-safety-enhancing-security-through-video-behavior-analytics/{/URL}
{Author}Nick Herbert, Forbes Councils Member{/Author}
{Image}https://imageio.forbes.com/specials-images/imageserve/66d1f8550dcd600f85169f18/0x0.jpg?format=jpg&height=600&width=1200&fit=bounds{/Image}
{Keywords}Innovation,/innovation,Innovation,/innovation,technology,standard{/Keywords}
{Source}POV{/Source}
{Thumb}{/Thumb}