Bottom Line:  Real-time analysis of remote video feeds is rapidly improving thanks to AI, increasing the accuracy of remote equipment and facility monitoring.

Agriculture, construction, oil & gas, utilities, and critical infrastructure all need to merge cybersecurity and physical security to adapt to an increasingly complex threatscape. What needs to be the top priority is improving the accuracy, insight, and speed of response to remote threats that AI-based video recognition systems provide. Machine learning techniques as part of a broader AI strategy are proving effective in identifying anomalies and threats in real-time using video, often correlating them back to cyber threats, which are often part of an orchestrated attack on remote facilities.

Identifying Anomalies In Real-Time

The future of remote security monitoring is being defined by the rapid advances in supervised, unsupervised, and reinforcement machine learning algorithms and their contributions to AI-based visual recognition systems. Video cameras capable of night vision, infrared, and thermal imaging act as sensors for these AI-based video recognition systems, providing 24/7 monitoring of remote equipment, sites, and the assets being protected there. One of the more interesting companies to watch in this area is Twenty20 Solutions, which is working to integrate machine learning with real-time video data feeds of remote sites and equipment to identify anomalies in real-time. Their SCADA-as-a-Service integrates sensors, gauges, and devices to provide data telemetry and real-time information. What’s noteworthy about this approach is the need many industries have to integrate cyber and physical security systems and get a 360-degree view of remote location threats. Integrating cyber and physical security by capitalizing on real-time monitoring across the oil and gas value chain delivers many operational benefits as well. The following graphic from PwC’s digital transformation in oil and gas illustrates the role of SCADA, real-time monitoring and integrated security in improving operational efficiencies:

HTTPS://WWW.PWC.COM/US/EN/INDUSTRIES/ENERGY/DIGITAL-IN-ENERGY.HTML

Supervisory control and data acquisition (SCADA) systems are used to monitor and control a plant or equipment in industries such as telecommunications, water and waste control, energy, oil and gas refining, and transportation. Research provider IMARC predicts the global SCADA market will be worth $26B by 2024, achieving a Compound Annual Growth Rate (CAGR) of 5.7% between 2019 and 2024. The following graphic is from their recent study, showing Oil & Gas, Manufacturing, Water & Waste Water, and Power or Utilities industries leading adoption.

HTTPS://WWW.IMARCGROUP.COM/GLOBAL-SCADA-MARKET

How AI and Machine Learning Are Defining The Future Of Remote Monitoring

AI and machine learning are technologies that excel at finding visually-based anomalies. Briefly, here are the three dominant types of machine learning algorithms current and future-generation AI-based video recognition systems are relying on:

• Supervised Learning Algorithms – Supervised machine learning algorithms excel at finding anomalies in images over time. They can do this by training on the data sets to identify correct images of objectives, so when an anomalous image appears, they can identify it. Agriculture, construction, oil & gas, and utilities companies rely on supervised machine learning algorithms to identify, track, and monitor vehicle, machinery, asset, and remote location use patterns. Construction companies benefit from these algorithms that protect remote sites and also anticipate potentially harmful work conditions for their production teams.  AI developers writing supervised machine learning algorithms for use in AI-based video recognition systems use Scikit-learn and Caret.
• Unsupervised Learning Algorithms – Algorithms in this category excel at discovering new patterns in images, which is invaluable in finding and reporting anomalies in real-time video streams. Oil & Gas companies rely on these types of algorithms for monitoring infrared and thermal data from remote equipment and assets. Popular tools for writing unsupervised machine learning algorithms include TensorFlow, PyTorch, and Keras. There’s a free deep learning tutorial from Stanford available online here.
• Reinforcement Learning Algorithms – Based on the concept of reinforcing outcomes, AI-based video recognition systems use these algorithms to course-correct how they identify and update known images. Remote construction, oil & gas, and utilities sites rely on the combination of real-time monitoring and reinforcement learning algorithms to constantly evaluate the condition of equipment and assets. The insights reinforcement learning algorithms provide helps to ensure the consistency, compliance levels, and safety of remote equipment located in diverse geographical locations stays at optimal levels. Google’s autonomous car project relies on reinforcement learning algorithms to navigate the test cities they are active in today, as does the Tesla self-drive function.

Machine Learning Algorithms Are A Perfect Fit For Visual Analysis

Machine learning algorithms are especially adept at finding patterns in data and drawing inferences from them. The ongoing research of AI-based vision recognition centers on ImageNet, a universally adopted standard that is used for evaluating the accuracy of AI-based computer vision interpretation. The computer scientists and data modeling experts who created the popular machine learning framework TensorFlow have found that convolutional neural network-based models are the most effective at accomplishing complex visual recognition tasks with the greatest accuracy. This area of research is heavily funded by venture capitalists who see the opportunity to provide advanced visual recognition systems across aerospace and defense, financial services, manufacturing, professional services, and medical applications of the technology.

AI-based visual recognition systems are capable of analyzing and interpreting a given image in seconds by iteratively sampling its data and using algorithms to classify, categorize, and create models representing the data sets in real-time.

Conclusion

Digitally-enabled remote cameras capable of night vision, infrared, and thermal imaging video streaming provide AI-based video recognition systems the data they need to protect remote locations, equipment, and assets. From an Industrial Internet of Things (IIoT) perspective, video cameras are quickly emerging as the smartest sensor of all, being able to monitor multiple properties of remote locations and assets in real-time and provide valuable insights to operations and security teams. Agriculture, construction, oil & gas, utilities, and critical infrastructure industries need to consider how their IIoT platforms can be integrated into a broader cybersecurity strategy to gain a 360-degree view of multifaceted threats being launched at their locations with increasing frequency. Closing the gaps between cyber and physical security is improving because of AI’s many contributions to identifying threats in real-time and stopping them.

Follow me on Twitter or LinkedIn. Check out my website. 

Original post: https://www.forbes.com/sites/louiscolumbus/2020/06/21/why-ai-is-the-future-of-remote-security-monitoring/