The Challenge
Rural crime has become an increasingly serious problem across the UK. Farm machinery thefts, livestock rustling, and property crimes cost agricultural businesses millions annually, with isolated locations making farms particularly vulnerable targets. Traditional security measures struggle in rural environments—CCTV systems designed for urban settings often fail in harsh outdoor conditions, and the sheer scale of farmland makes comprehensive monitoring impractical.
Farmers and rural residents need security solutions that work reliably in remote locations, can distinguish between normal farm activity and genuine threats, and provide immediate alerts when suspicious activity occurs. Existing systems generate too many false positives, require constant internet connectivity, or simply aren't designed for agricultural environments.
Our Approach
Working with Farmstream, New Gradient developed an intelligent security system that combines multiple AI capabilities into a practical, rural-ready solution. The system builds upon Farmstream's existing camera infrastructure, adding sophisticated detection capabilities without requiring wholesale hardware replacement.
At the core of our solution is a custom object detection model trained specifically for agricultural environments. Unlike generic security systems, our models understand the context of farm activity—they recognise livestock, distinguish between agricultural vehicles and unfamiliar cars, and learn the patterns of normal farm operations. This contextual awareness dramatically reduces false alarms while ensuring genuine threats don't go unnoticed.
We developed a novel Automatic Number Plate Recognition (ANPR) system designed for the challenging conditions found on farms. Traditional ANPR requires vehicles to pass at specific angles and distances; our system handles oblique viewing angles, varying lighting conditions, and partially obscured plates—common scenarios on farm tracks and rural roads. The system maintains a database of recognised vehicles, alerting farmers only when unfamiliar number plates are detected.
The complete solution runs on edge devices at each camera location, processing video locally rather than streaming to cloud servers. This approach works reliably even with limited rural internet connectivity, reduces latency to enable real-time alerts, and addresses privacy concerns by keeping footage on-premise unless explicitly shared.
The Outcome
The deployed system provides farmers with genuine peace of mind through reliable, intelligent monitoring. Real-time alerts notify users instantly when unfamiliar vehicles enter their property or when unusual activity is detected, delivered directly to mobile devices regardless of where the farmer is working.
The edge-based architecture proves particularly valuable in rural settings where internet connectivity can be unreliable or expensive. Cameras continue monitoring and detecting even when offline, queuing alerts for delivery when connectivity returns. This resilience is essential for remote locations where traditional cloud-dependent systems would be impractical.
Beyond security, the system provides operational benefits. Farmers use the livestock detection capabilities to monitor animal welfare, receiving alerts if animals appear distressed or have escaped from fields. The historical data helps track activity patterns, supporting both security planning and general farm management decisions.