Alpha Secure Solar Farms:
Case Study

The challenge

The goal to be achieved is to detect possible intruders in real-time with maximum precision, day or night, with a minimum number of false positives. The challenge here is to prove we can greatly outperform the incumbent on-camera analytics, which have been proved very ineffective against intruders entering the perimeter crawling or rolling on the ground at medium distances from the camera, particularly at night. Additionally, the on-camera analytics have generated a huge quantity of false positives, stemming from environmental factors (wind, rain), and the movement of birds and other animals appearing in the scene.

The attacks on solar farms have, over time, been perfected to minimise the risk of detection. Typically, gangs now enter the farms in the middle of the night from the farthest distance away from any existing camera. The intruders will actually crawl or even roll while facing any existing cameras to minimise the surface area where motion can potentially be detected. This is not one of those cases where forensic searches are of any value; the intruders are wearing hoodies, so there’s no possibility of later using the recorded video to recognise suspects.

The solution: IntelexVision iSentry system

IntelexVision’s iSentry platform addressed all the above challenges. Firstly, it very quickly learns the scene (few seconds) and is able to hugely reduce the false positives generated by environmental factors. It is also able to detect movement at distances previously undetectable, particularly at the limit of what the existing camera manufacturers informed as “maximum infra-red illumination range”. Last, but not least, people crawling or rolling across the no-man land were identified over 99.5% of the time.

The capability to clearly classify all kinds of moving objects during the day (separating people from birds, animals, tractors, vehicles, etc.) also helps reduce the quantity of false positives reaching the operators, as the Intelex Rules Engine will filter out these so-called “intrusion” events.

Finally, if events meeting pre-determined conditions of severity are identified, snapshots and short video clips of the event are then sent to the operator for immediate analysis. At this point, the operator will determine whether the event should be dismissed (for example he identifies a maintenance crew or a guard executing his patrolling) or should be elevated to an alarm condition.

The solution was deployed running the iSentry Licenses on an edge device on the solar farms – usually an i7 NUC with 32 Gigs of RAM, OS Storage of 240GB SSD, DB Storage of 1TB SSD, OS SW Win 11 Pro, one NUC capable of processing 15-20 RTSP video streams coming from the already installed cameras on each farm.

Then, through a low bandwidth redundant internet connection, all alerts were then sent to the Alpha Secure monitoring centre, situated in the city of Sao Paulo.

The iSentry platform additionally provides sophisticated Business Intelligence tools to inform how operators in the control room are handling incidents (time to address an alarm and to actually acknowledge the incident), shows which cameras are generating most alerts (in a 24 hour graph showing alert volume per hour) or how many alerts are being processed and automatically dismissed by the rules engine and how many are processed by the operator. This data is essential to the continuous improvement process of the monitoring centre.

The outcome

“IntelexVision´s iSentry video analytics platform has been successfully deployed on more than one hundred solar farms that we are monitoring for intrusion, generating a huge improvement in overall efficiency, both due to more true detections, especially in harsh conditions (nights, high wind or rain, etc), as well as a generating a huge reduction in false positives, thereby allowing our operators to have the necessary time to analyse real incoming alerts. We have changed our operation from trying to identify intrusions on motion detection alerts to having a user interface that actually helps the operator visualize alerts showing bounding boxes on the subject generating the alert with an additional short video clip of the pre and post alarm event, all in one single UI. As important to us is IntelexVision´s permanent availability to help us with any request for system analysis, opportunities to improve the system calibration to generate even better detections, as well as the BI and forensic search offered at no additional cost, which allows us to send weekly reports of system performance to our end users. We are planning to deploy this solution in many other solar farms we are now adding to our customer base.”

Antonio Marcos
Commercial Director
Alpha Secure