Video recognition software

In Texas’ sprawling and prolific Eagle Ford play, an active major oil and gas company conducts drilling operations spanning more than 200 000 acres. As part of its SCADA and Local Area Network (LAN) systems, communications towers with Axis security cameras were installed and integrated with existing rule-based analytics software.

Although this surveillance network is generally comparable to other companies’ E&P operations, too many flaws became apparent in this ‘typical’ surveillance set-up. After evaluating the situation, the company realised it needed a more innovative solution leveraging the latest technology.

Recognising the problems

One of the biggest problems with the existing combination video surveillance/analytics software was that it perceived virtually any movement as hostile and sent video alerts for anything from wind to animals. Yet, an even larger issue was the system’s inability to distinguish normal from abnormal activity, with all the false alarms culminating in a system of no more value than not having one installed in the first place. More than just a non-productive inconvenience, these false alarms required personnel to re-set the software after each ‘non-event.”

As a result, the E&P major called in a vendor to eliminate these problems along with some associated issues. For example, the company wanted to know if downtime could be improved, if vandalism could be reduced and if safety could be enhanced since usually only one employee was on-site. Additional concerns involved integrating new software with current LAN resources, how to better handle video alerts and, above all, how to better visually pinpoint abnormal behavior at the sites as contrasted with normal behavior.

Getting technologically current

Meanwhile, advances in video surveillance had actually been around a while and, with certain innovative developments, would prove to be the solution for this oil company. Technological surveillance’s history generally begins with cameras at fixed installations recording what appeared in their field of vision. Then video analytic software changed the playing field on a rule-based basis, which sent alarms when ‘lines’ were crossed and now a Houston company has developed behavioral analytic software that takes video surveillance to the next level.

What the new generation of video surveillance comes down to is essentially real-time situational awareness as contrasted with the old rule-based video analytics. The latter, according to a knowledgeable source, “not only takes a tremendous amount of investment and time to configure but is just as cumbersome in set-up and writing the system’s rules.” For those not familiar with rule-based analytics, a line could literally be drawn on a sidewalk and if someone or any object as inconsequential as a blown piece of paper crosses the line, an alarm is activated and the software must be repeatedly re-set.

Case study – oil company

For this major oil company working in the Eagle Ford play, the new software was installed at the eight remote tower sites with relative ease. Initially, security cameras had been installed at these sites, only to encounter the problems that have been the bane of older legacy video analytic systems. With the new enhanced installation, however, there was actually a fairly quick dual improvement. 

Real-time situational awareness translated into delivering real-time alerts about real threats. In doing so, it ‘learned’ to disregard the wide-ranging array of everyday movements that have traditionally caused so many false alarms. And, at the same time, it continually taught itself what was ‘normal’ and what was not. Objects entering its field of vision were analysed based on appearance, classification and interaction with the environment. It also records time stamps. Not only does this learning occur non-stop, the more it learns the more fine-tuned its memory becomes.

With the new software, which integrates with most video management systems, security personnel began receiving bonafide real-time alerts and not the plethora of bogus alerts to which they had become accustomed. And because the software’s self-learning memory taught itself to recognise only real threats, it has freed personnel from being bogged down in pre-programming/re-programming. Additionally, since the system is not rule-based, it can be quickly installed.

The company has also experienced other value-adds, including the fact that fewer personnel are required for set-up and ongoing operations and maintenance. Also, different departments within the company, such as operations, physical security and HSE are kept ‘in the loop,’ receiving actionable alerts so that they are apprised of possible threats nearby. The company also reported less theft, vandalism and other safety problems.

Areas of heavy E&P activity such as Eagle Ford have heightened the need for technological advances that exceed both standard videos surveillance and now outdated analytics software. Video recognition software became that solution for this oil company and points the way for even wider usage.

 Written by Chuck Drobny, GlobaLogix, USA.

Edited by Callum O'Reilly

Read the article online at: https://www.hydrocarbonengineering.com/special-reports/17022014/video_recognition_software_oil_company/