Aurora
– Power
of Gen AI

Pioneering Generative AI in security industry

Aurora is an innovative, end-to-end trained large multimodal model that merges a vision encoder with language understanding capabilities. It achieves remarkable conversational abilities reminiscent of GPT (Generative Pre-trained Transformer) while setting new benchmarks for state-of-the-art accuracy in video processing and contextualising alerts and images.

Aurora acts like a virtual operator, allowing you to ask chat-like questions to identify threats that aren’t immediately visible – from weapons on a scene to potential fire hazards, just taking these examples.

How it works?

1. POSE THE QUESTION

The operator or user can pose chat-like questions to Aurora to extract more specific information from the alert image. For example, whether any weapons are visible on the scene, or if there’s a fire or safety violation.

2. ANALYSE THE IMAGE BY LARGE MULTIMODAL MODEL

Aurora uses its end-to-end trained large multimodal model capabilities, combining a vision encoder with language understanding. It analyses the content of a single alert image and attempts to contextualise it, helping operators identify unusual aspects that might not be immediately obvious.

3. GET THE CONTEXTUALISED CONTENT

Once the image has been analysed, Aurora provides contextual information about potential threats or unusual activities visible in the scene. It adds a layer of verification especially in cases where the alert is unclear or not automatically categorised.

4. PROVIDE THE ANSWER TO USER

Aurora functions as a virtual operator, providing answers and detailed insights in complex scenarios. It enhances the operator’s understanding of the situation and boosts confidence by accurately contextualising and explaining the scene.

5. ACT AS VIRTUAL OPERATOR

Finally, Aurora acts like a virtual operator within the security system, offering decision support by escalating alarms and aiding in real-time security management, ensuring prompt and efficient responses to potential security threats.

Technical partners:

Integration partners:

Core capabilities

Using a large multimodal model like Aurora in security applications significantly enhances situational awareness and decision-making. By integrating vision and language understanding, Aurora analyses visual data and provide contextual insights, leading to a deeper understanding of events. This model’s accuracy and real-time processing capabilities reduce false alarms and ensure prompt responses to real threats. Its extensive training on diverse datasets improves reliability across various scenarios, critical for maintaining security in different environments.

Enhanced Situational Awareness

Accuracy and Reliability

Real-time Decision Support

Scalability and Adaptability

Reduction in Operator Fatigue

Enhanced Situational Awareness

Aurora leverages its integration of advanced vision and language processing capabilities to offer a more nuanced interpretation of visual data.

This technology goes beyond traditional image recognition by understanding the context within which activities occur, identifying subtle nuances that might indicate irregular or suspicious behaviour.

This level of analysis enables security personnel to gain a holistic view of the situation, providing them with critical insights into not only what is happening but also the potential security implications.

Such comprehensive situational awareness is crucial in preempting threats and managing security more proactively.

Accuracy and Reliability

Aurora is built upon a foundation of extensive datasets that encompass a wide range of scenarios, which is essential for training the model to recognize and interpret diverse and complex security incidents accurately.

This robust training regimen enhances the model’s ability to deliver reliable alerts and minimise the occurrence of false positives, a common issue in less sophisticated systems.

The high accuracy and reliability of Aurora ensure that security teams can trust the alerts they receive, allowing them to respond more effectively to actual threats without wasting resources on false alarms.

Real-time Decision Support

Aurora’s real-time processing capabilities are a critical asset in security settings where decisions must be made quickly to mitigate potential threats.

By analysing data on-the-fly, Aurora provides immediate feedback and guidance to security operators, enabling them to make informed decisions swiftly.

This real-time support is invaluable during urgent situations, helping to streamline the decision-making process and ensuring that security personnel have access to the most current information when it matters most.

Scalability and Adaptability

The scalability of Aurora means it can be effectively implemented in various settings, from small, private venues to large, complex public spaces.

This adaptability is further enhanced by its capacity to learn from new data, allowing the model to adjust to emerging security challenges and continuously improve its effectiveness.

Aurora’s flexible design ensures that it can evolve with changing security needs and integrate seamlessly into existing security infrastructures, making it a versatile solution adaptable to any security requirement.

Reduction in Operator Fatigue

One of the significant advantages of Aurora is its ability to automate the initial stages of threat detection and analysis, which can be tedious and error-prone when performed manually.

This automation significantly reduces the cognitive load and potential fatigue among operators, who otherwise would need to monitor multiple feeds and inputs continuously.

By handling the routine detection tasks, Aurora allows operators to focus their attention on critical thinking and decision-making, thereby enhancing their efficiency and reducing the risk of overlooking potential threats.

Reliable Information – Enhanced Contextual Analysis

Utilising a sophisticated array of large multimodal models, Aurora significantly enhances security operations by providing in-depth analysis of surveillance footage. Aurora’s advanced capabilities stem from its integration of vision encoders and language processing, allowing it to recognise and interpret complex scenes within a wide range of environments. This powerful system excels even in challenging scenarios such as poor lighting or obscured views.

Aurora’s neural networks are continually refined to include specialised capabilities tailored to unique security needs. Each update to Aurora’s system incorporates the latest advancements in artificial intelligence, maintaining its edge in threat detection and situational awareness. By consistently evolving, Aurora ensures it delivers reliable and actionable information to maintain safety and compliance in various settings.

Aurora – Your New Virtual Operator

Aurora also functions as a Virtual Operator, an innovative feature that significantly enhances interaction between security personnel and the surveillance system. As a Virtual Operator, Aurora can process natural language queries from users, allowing them to ask specific questions and receive real-time, contextually relevant information about any aspect of the security environment. This capability transforms Aurora from a passive monitoring tool into an active participant in security management, offering detailed explanations, escalating alerts when necessary, and providing guidance during complex security situations. By enabling more intuitive and interactive surveillance operations, Aurora facilitates a more dynamic and responsive security management process, ensuring that operators have immediate access to crucial insights exactly when needed.

Flexible Architecture

Aurora enhances iSentry capabilities and inherits its architecture. It is designed to work with different infrastructure environments, maximising the value of existing hardware investments and scaling flexibly and efficiently as possible, keeping costs to a minimum without losing performance.

Centralised Architecture

Distributed Architecture

Complete Edge Architecture

Partial Edge Architecture

Centralised Architecture

Aurora and iSentry Centralised Architecture has the advantage of low complexity taking advantage of economies of scale. In the case of a network containing many cameras, it could require extra bandwidth to ensure centralisation of all video in a single central location.

Distributed Architecture

Complete Edge Architecture

Partial Edge Architecture

iSentry Holistic Solution

Aurora is part of the iSentry Premium and iSentry Elite licences.
It does not appear on its own.

COMPETENCE IN AI

Aurora stands as a pioneering solution in the security market, being the first to leverage a large multimodal model to enrich and contextualise alert information derived from iSentry. This groundbreaking approach allows Aurora to offer unprecedented situational awareness by intelligently analysing and interpreting data from various security feeds.

The system’s capability to integrate visual data with linguistic insights enables it to provide a more nuanced understanding of each alert, transforming raw data into actionable intelligence.

COMPETENCE IN AI

iSentry Docker Solution
1. Decoding video
2. Processing video with core Al and producing alerts
3. Enriched alerts using Deep Learning Al
4. Enrich with Aurora
5. Decision making with the Rules Engine