Biobiosecurity and analytics system at a poultry farm

The large agricultural enterprise, the "Sovetskaya" poultry farm, faced a systemic problem: the facility's biosecurity was virtually nonexistent.

Project background


The large agricultural enterprise, the "Sovetskaya" poultry farm, faced a systemic problem: the facility's biosecurity was virtually nonexistent. The site lacked access control and employee photo identification, monitoring of workwear and PPE usage, automated control of passage through disinfection barriers, a centralized video surveillance system for workshops, warehouses, and the perimeter, control of technological processes (feed loading, manure unloading, egg counting), as well as an early detection system for fires and smoke.
 

Each of these gaps posed a direct threat to the enterprise's biosecurity and a risk of mass poultry mortality. A comprehensive solution was required that would cover all aspects of security and control, eliminating the "human factor" and ensuring centralized management.
 

Implementation progress


Specialists from OSMOVIEW LLC developed a comprehensive security system based on the OSMOCV platform. The integrator provided equipment supply, installation, network cabling, and access control system (ACS) setup; OSMOVIEW LLC provided the platform, neural services, and scenario analytics.
 

Key components of the solution:

  1. Local Area Network — a fault-tolerant structured cabling system with fiber optic backbones and Wi-Fi bridges for remote facilities was created, ensuring data transmission from 240+ IP cameras, access terminals, and controllers.
  2. Access control and photo identification system — biometric Hikvision terminals were installed at key points (face and card recognition was performed by the customer's existing ACS, which sent data to the platform for further processing), while Sigur E510 controllers ensure access segregation, integration with the video surveillance system, and violation logging.
  3. Video analytics for biosecurity — the OSMOCV platform with neural services solves a wide range of tasks in real time:
  • detection of a person in "clean" and "dirty" zones;
  • movement of people and vehicles from the "dirty zone" to the "clean zone";
  • control of workwear and disposable suit (PPE) usage;
  • recognition of vehicle license plates at disinfection barriers;
  • monitoring of feed loading and manure unloading with automatic PTZ camera tour activation;
  • report generation and analysis of egg counts for various periods (day, week, etc.);
  • detection of open flames and smoke in warehouse areas and the boiler room.
  1. Modernization of disinfection barriers — automatic barriers, passage time timers, disinfectant level sensors, traffic lights, and a license plate recognition system were installed for "clean" and "dirty" entrances. Violation of the passage protocol prevents entry onto the enterprise's territory.
  2. Life support systems monitoring — using Laurent-2 network controllers, collection of discrete signals from SCADA equipment was configured: power supply (no signal), fire system, water pumps (on or off), liquid level in disinfection barriers and life support systems. Information is displayed on operator dashboards in a unified situation center.
  3. Situation center and incident management — all events from subsystems are aggregated, normalized, and enriched with context. Automatic incident routing, SLA monitoring for resolution, and notifications via Telegram and email have been implemented.
     

Scenario analytics implemented on the platform:

  • vehicle parking (stopping) near the perimeter;
  • detection and recognition of special workwear and PPE;
  • perimeter crossing by a person or animal;
  • license plate recognition with verification against a database of authorized plates;
  • automated video monitoring of feed loading and manure unloading;
  • person detection in designated zones with schedule consideration;
  • detection of open flames and smoke;
  • vehicle type recognition;
     

Parallel processing architecture

A key feature of the project is the architecture built on five Vimark servers operating in parallel mode. Each server hosts neural services and analytics modules that process video streams from the cameras assigned to it. Key information about events and incidents is aggregated on a central server, which serves as the single point of data collection and management for the situation center.
 

Achieved results


  • Full biosecurity control has been ensured — all key points have been automated: the checkpoint, workshops, disinfection barriers, warehouses, and perimeter. Biosecurity violations have been eliminated.
  • Security costs have been reduced by 20% — the system has taken over routine monitoring, reducing the workload on the security service.
  • 73% of biosecurity violations are recorded automatically — the human factor has been eliminated from the control process, and violations are detected in real time.
  • 80% of violations are resolved within 24 hours — thanks to prompt alerts and automatic incident routing.
  • PPE control has been automated — the system records the absence of PPE and leaving the premises in workwear in real time.
  • Disinfection barrier control has been ensured — violations of the passage regulations (time, solution level, unauthorized number) have been eliminated.
  • Staff discipline has been improved — the system notifies about every violation (Telegram, email, pop-up windows in the interface), which reduces the number of incidents.
  • Monitoring of technological processes has been automated — feed loading and manure unloading are controlled with automatic PTZ camera activation.
  • Accurate product accounting has been implemented — report generation for any period.
  • Early fire detection has been ensured — fire and smoke detection transmits an alarm to the operator in real time, minimizing the consequences.
  • Detection of life support system violations in workshops — monitoring of power supply, pumps, and liquid level in disinfection barriers.
  • A unified situation center has been created — all incidents are aggregated, routed, and controlled according to SLA, with notifications via Telegram and email.