Ron Kimmel, Chief Scientific Officer, Lumana explains how AI and edge analytics are pioneering innovations for security.

AI innovations

The traditional approach to video security has long been reactive, with security teams monitoring live feeds and poring over footage after incidents occur.

With the emergence of AI and edge analytics, we’re witnessing a fundamental shift in how modern organizations approach security.

The convergence of these technologies is changing how video analysis and surveillance are conducted, addressing long-standing challenges like latency, data privacy, scalability and adaptability.

This article looks at some of the latest AI and edge analytics innovations, explores the significance of hybrid architecture for video analysis and shares its potential for addressing industry challenges.

A paradigm shift

Technological advancements are always moving the security industry forward, making it more efficient and better adapted to modern threats.

In recent decades, as organizations turned to live streaming to monitor their facilities, video analysis systems relied heavily on centralized cloud infrastructure for the storage and processing capacity of large volumes of video data.

This infrastructure enabled advanced analytics while also allowing remote user access.

While effective for many use cases, these cloud systems often struggled with latency issues and bandwidth constraints.

Due to their reliance on centralized data storage, they also raised concerns over data privacy.

Edge analytics represents a key advancement in security technology.

Rather than sending all data to centralized servers for processing, edge devices conduct analysis closer to the data source.

This technological advancement has emerged as a game-changer for the industry.

By deploying advanced AI models on edge devices, organizations can:

• Achieve real-time insights: Immediate response times are critical in the security industry, where seconds can make a difference in preventing incidents
• Preserve data privacy: Processing sensitive visual data locally minimizes the risks associated with transmitting data to centralized servers
• Optimize bandwidth usage: By analyzing data at the source, only actionable insights are transmitted to the cloud, reducing network congestion and associated costs

Key innovations in AI and edge analytics for security

AI at the edge signifies a transformative shift from traditional data analysis models.

Advances in computation, AI, connectivity and frameworks have made it possible for edge analytics to achieve the following:

1. Hybrid architecture for video analysis

Modern security systems combine edge and cloud capabilities, creating a robust infrastructure where edge devices handle real-time processing while the cloud manages storage, advanced analytics and model training.

This distributed approach ensures system resilience and enables continuous learning.

2. Comprehensive threat detection

With parallel processing of threat indicators, AI-powered edge systems can now detect a wide range of security threats, including:

• Objects from unconventional angles
• Behavioral anomalies and pattern deviations
• Multiple object classes (faces, license plates, weapons)

3. Enhanced interoperability

Edge analytics enables seamless integration with existing security infrastructure.

Advanced systems can communicate with Internet of Things (IoT) devices, access control systems and alarms to provide a cohesive security ecosystem.

Edge analytics ensures that the data from various sources is analyzed and utilized effectively.

4. Energy efficiency

Bypassing the need to transmit video data to cloud servers, modern edge devices are optimized for low power consumption.

This reduced energy usage makes it ideal for remote deployments with limited power resources, as well as organizations who want to minimize their energy footprint.

Overcoming challenges in the security industry

AI-powered edge systems are particularly well-poised to solve the following challenges with existing security systems:

Latency and response time

When it comes to security, a delayed response can lead to catastrophic outcomes.

With sub-second response times, edge analytics speeds up threat detection and mitigation.

It eliminates the delays associated with transmitting data to centralized systems for processing.

Scalability and cost efficiency

Edge devices eliminate the need for constant cloud connectivity, significantly reducing operational costs.

Additionally, these systems are highly scalable, allowing organizations to expand their security networks without overhauling their infrastructure or implementing complex network configurations.

Data privacy and compliance

With increasing regulatory scrutiny around data privacy (e.g., GDPR, CCPA), processing data locally on edge devices reduces the risks associated with data transmission and centralized storage.

Organizations can maintain compliance while safeguarding sensitive information.

With these challenges adequately addressed, organizations can greatly benefit from enhanced situational awareness and can react quickly and proactively to any threats.

Example use case: Paris Olympics 2024

In the lead-up to the Paris Olympics in 2024, heightened security risk necessitated an increase in surveillance efforts.

With estimates of ten million spectators across several weeks of sporting events, powerful security cameras were deployed at various Olympic sites across the city, with AI-assisted software to detect trouble.

The advanced AI powering these systems could flag certain events such as movement, changes in crowd size, abandoned objects and/or the presence or use of weapons.

For instance, the system could immediately detect an event like a crowd surging toward a gate or a person leaving a backpack on a crowded street corner and immediately alert security personnel.

This real-time detection capability allowed for swifter, more efficient responses, making these systems far more effective than their predecessors in managing and mitigating security risks.

The road ahead

We are only beginning to scratch the surface of what capabilities will be possible with new advancements in AI and edge analytics.

The following are three examples of what we can expect going forward:

Advancements in AI models

Future innovations in AI for edge devices will focus on two key areas: lightweight models optimized for resource-constrained devices and federated learning that enables collaborative data sharing while maintaining privacy.

Edge-to-cloud synergy

5G networks will enable faster data transfer between edge and cloud systems, enhancing overall security capabilities.

Working together, edge and cloud systems will ensure immediate responses at the edge while leveraging the cloud for long-term data analysis, scalability and advanced machine learning models.

Expanded use cases

Beyond traditional security applications, AI and edge analytics will likely play a pivotal role in:

• Smart cities: Monitoring traffic, crowd density and environmental conditions
• Industrial safety: Detecting equipment malfunctions and ensuring compliance with safety protocols
• Healthcare: Monitoring patient movements and identifying potential hazards in real time

The future outlook for security

The convergence of AI and edge analytics represents a fundamental shift in security technology.

These innovations are generating more intelligent, responsive and adaptive security ecosystems while addressing data privacy and data protection concerns.

The transformation of security through AI and edge analytics is not just about deploying new technology – it’s about reimagining how we approach security in an increasingly complex landscape.

Organizations that embrace these innovations while staying mindful of their associated challenges will set new standards for security effectiveness and efficiency.

By harnessing AI and edge analytics, today’s organizations can tap into the power of technology to shape a safer, smarter world.

About the author

Ron Kimmel is the Chief Scientific Officer at Lumana, an AI video security system designed to automate monitoring, identify actionable insights and accelerate incident response.

His expertise spans image processing and analysis, medical imaging, biometrics, computer vision, machine learning, 3D recognition, geometry processing and augmented and virtual reality.

In addition to his role at Lumana, Ron is a Professor of Computer Science and Electrical Engineering at the Technion (Israel Institute of Technology).

This article was originally published in the special February Influencers Edition of Security Journal Americas. To read your FREE digital edition, click here.