3DiVi unveils Omni Agent 3.0 to boost video analytics efficiency

3DiVi has introduced Omni Agent 3.0, the latest version of its platform for intelligent video analytics. The update centres on a completely redesigned architecture, expanded detection capabilities, and significant improvements in efficiency. These enhancements are designed to help businesses improve real-time monitoring, reduce operating costs, and enable more tailored use of video analytics across a range of industries.

Architecture overhaul

The new release features a reengineered pipeline management system.

This gives users the option to create custom video analytics scenarios without reinstalling or reconfiguring the entire agent. Organisations can combine detection modules and tailor alerts to match specific operational requirements. This flexibility enables more precise and contextually relevant monitoring.

Detection logic can be updated with no manual intervention, as updates to weights and templates can be delivered directly over the network. This is intended to keep recognition models current as environmental conditions or business processes change over time.

Expanded detection

Omni Agent 3.0 brings an expanded set of detection capabilities, including specialised detectors for mobile phones and an advanced Unusual Activity Detection module.

The new angle-aware detection setting allows users to improve accuracy for cameras deployed in various positions, such as vertical or overview orientations. This is especially relevant in settings like retail aisles or industrial walkways.

Retrained recognition models have enhanced performance in challenging lighting conditions, including low-light indoor environments and high-contrast outdoor locations. These updates are being used in industrial settings where accurate people and event detection underpins worker safety, perimeter security, and operational oversight.

Additional configuration controls allow users to define object sizes for region-of-interest filters, set event timeouts, and create line-crossing triggers. These tools reduce false positives and increase the reliability of real alerts for operators.

Lowered hardware demands

With Omni Agent 3.0, analytics scenarios can be applied on a per-camera basis. This offers more control over the allocation of compute resources. Each camera can have its own detection parameters, zones, and triggers, letting businesses focus processing power only where it is most needed. Even when analytics is selective, full-frame video is preserved for relevant events to ensure thorough evidence for investigations.

An external API now permits the remote starting and stopping of analytics scenarios, supporting automated workflows and integration with broader systems.

The platform supports TensorRT 8x acceleration on Linux, increasing inference speed for edge devices while reducing hardware requirements.

Windows systems continue to benefit from CUDA acceleration. These changes contribute to extending device lifecycles and reducing the cost of scaling up advanced analytics deployments.

Web-based configuration

The configuration process has moved entirely to a web interface, eliminating the need for manual file editing or command-line operations. Operators and integrators can set up or modify complex analytics pipelines more quickly and with less technical overhead.

The interface includes a centralised workspace with licensing and resource panels, preview controls for object rendering, and real-time monitoring of system loads. Tools are available for creating, editing, or cloning detection templates, allowing for rapid deployment of new analytics scenarios.

Business applications

The changes to Omni Agent 3.0 seek to support a broad range of business needs, including security, compliance, retail analytics, industrial safety, and smart city monitoring. By reducing both hardware requirements and operator workloads, the platform aims to make large-scale, real-time deployments more accessible in environments with limited technical resources.

"The release of Omni Agent 3.0 demonstrates our commitment to making real-time video intelligence more accessible, cost-effective, and configurable for enterprises of all sizes," said Ivan Oshchepkov, Chief Executive Officer, 3DiVi.