An integrated and customised Video Analytics solution for the real-time events detection

AiVu-Smart Modules is a modular Video Analytics solution developed by Aitek to process live and recorded image from video-surveillance cameras.

Whether transportation infrastructures, urban areas, points of sales, industrial sites or other are involved, security personnel require reliable solutions capable of detecting any threat in real time. This is why video analytics have become an essential tool for guaranteeing immediate responses in case of events representing possible dangers to people, goods and infrastructures.

The AiVu-Smart Modules suite integrates existing CCTV systems for the real-time detection of a wide range of events focusing on people, vehicles and static objects. Our software modules are also designed to apply post-event analytics to recordings and to provide useful data for statistical and business intelligence analysis.

The employed image processing algorithms automatically generate notifications whenever events of interest occur, configured based on the needs of the surveillance system manager. The detection is highly reliable and accurate and any redundant or false alarms are automatically filtered out.

AiVu-Smart Modules solutions are scalable and adaptable, allowing users to add new features simply by installing additional licenses when needed.

Features

Modular video analytics platform for the design of solutions tailored to the Cusotmers’ needs
Easily expandable with new functionalities
User-friendly web-based configuration interface that does not require specific know-how
Increased operating efficiency and reduced response times in case of emergencies
Post analysis on recordings to detect events without having to browse hours of video streams
Automatic forwarding of alarms and relevant recordings to the control centre and to mobile devices
Default values and/or advised interval ranges for each parameter
Deep learning, data analysis and design of neural networks for even more reliable video analytics
Integration in third-party systems (i.e. railway traffic control systems, traffic management, SCADA , etc)

Easy and quick video analytics configuration

The front end is a web-based interface, supported by mainstream browsers and therefore easily accessible from workstations equipped with different OSs (Windows, Linux), which allows to draw virtual sensors (“rules”) on an image or parts of it.

The number and type of sensors on the cameras are easy to modify. Multiple sensors can be configured on a single camera to simultaneously detect different events without affecting processing performance.

Setting up video analytics on a video camera is extremely simple: one just has to select the rule corresponding to the event to be detected and to define the image area to which the analytics must be applied by drawing the virtual sensor using the mouse.

Furthermore, default values and/or suggested ranges are available for all parameters and thresholds, allowing even video processing novices to perform analytics configuration with no need for complex measurements or evaluations.

If a video camera points to a different location, video analytics can be restored simply by adjusting the position of the virtual sensors on the image.

Post-event analytics on recordings

The video analytics software modules can also be applied to recordings, allowing to quickly and automatically reconstruct events of interest, without having to examine hours of recordings. Events can be quickly detected just by drawing the virtual sensors on the images: post-event analytics are performed at the highest possible speed, allowing to process hours of recordings in a few minutes!

Real-time alarm notification

All notifications generated by sensors, with the corresponding images, can be automatically sent to a control center provided with Aitek’s AiVu-VMS video management portal or to third-party supervisory systems, using the AiVu format or Onvif specifications. The control center automatically receives alarm notifications from video cameras any time unpredicted events occur, allowing the security personnel to view the recordings associated with the event and trigger the appropriate response procedures.

Deep learning: the evolution of video analytics

Deep learning is based on training highly sophisticated neural networks to achieve the highest reliability in processing images and video streams.

The performance of neural networks improves as they are fed more examples from which to learn, allowing the algorithm to dynamically adapt to new conditions. In case of changes to the application scenario or of incorrect detections, a network can be retrained to correct any inaccuracy.

Deep learning provides numerous advantages. Using neural networks provides robustness to rapid variations in the monitored scene such as changes in weather, lighting or video camera orientation.

Furthermore, algorithms can process in real time every single image with no need to employ a reference model of the scene and can detect with extreme reliability even partially overlapping or occluded objects or objects which have long been abandoned.

Strengths

Aitek expertise in data analysis for neural network design
Development of complex video analytics solutions with no need to develop complex algorithms
Detection of individual objects even when overlapped or partially occluded
Classification of objects comparable in shape and size (for example, bicycles/motorcycles)
Detection of motionless objects in the monitored scene (long-term tracking)
Audio stream processing for sound detection applications

The AiVu-Smart Modules algortihms

AiVu-Smart Modules-Calibration

This module provides camera calibration which allows to transform image coordinates into real-world coordinates (and vice versa) and compute distances, speeds, etc. The calibration can be applied to fixed cameras, in which case the resulting calibration is only valid for a specific camera, or to PTZ cameras.

AiVu-Smart Modules-Count

This rule counts all objects detected in the virtual sensor drawn on the image. It is ideal for counting vehicles travelling along a road lane or people/vehicles at access gates.

AiVu-Smart Modules-Crowd

This rule detects overcrowding within a configured area by setting minimum crowd occupancy and maximum stationing time. It is ideal for the surveillance of squares, public spaces, railway platforms and stations, entry and exit gates, waiting areas, ticket offices, etc.

AiVu-Smart Modules-Danger

This module detects the presence of people, vehicles and objects in a sensitive area, generating an alarm after a configurable time threshold. It is ideal for monitoring operational areas, warehouses and depots and for detecting potentially dangerous situations in railway scenarios such as objects on the rails, yellow line crossing, etc.

AiVu-Smart Modules-Debris

This module developed specifically for roadway scenarios detects motionless objects in a sensitive area which can be classified as cargo lost by vehicles in transit.

AiVu-Smart Modules-Direction

This module detects objects moving along a reference direction, defined during module configuration. An alarm notification is generated any time an object crosses the monitored area moving along the configured direction. It can optionally ignore pedestrian. Ideal for detecting wrong-way vehicles or people crossing unidirectional gates in forbidden ways.

AiVu-Smart Modules-Heatmap

This module generates a heatmap of objects transits in the monitored scene, i.e. it provides a graphical representation, using pseudo colors, of areas most frequently occupied by objects. Ideal for the analysis of vehicle flows along roadways or for point of sale in-store analytics.

AiVu-Smart Modules-Interdistance

This module computes a set of statistics on objects crossing the monitored scene: absolute count, hourly count (transits per hour), distance between objects.

AiVu-Smart Modules-Intrusion

This module generates an alarm when an object (for example a vehicle or a person) is detected in the virtual sensor drawn on the image for longer than a configurable time threshold. It is ideal for perimeter protection and detecting trespassing over gates and fences in industrial sites, port terminals, warehouses and depots, military sites, etc.

AiVu-Smart Modules-Position

This rule detects vehicles and/or people in forbidden areas, emergency lanes, etc. An alarm signal is generated each time people or vehicles are detected within a configured sensor area (with no time threshold). This module is best suited to the surveillance of non-entry or restricted areas, such as parking-lots, worksites, depots, warehouses, industrial sites, etc.

AiVu-Smart Modules-Queue

An alarm notification is generated when the monitored area (for example a road lane) is occupied by vehicles over a set threshold percentage. Ideal for monitoring traffic flows and detecting queues and accidents.

AiVu-Smart Modules-Smoke

This module employs a neural network to detect smoke and fog, even if not particularly dense, in outdoor or indoor areas (for example in tunnels or industrial sites) to provide early warnings in case of fires.

AiVu-Smart Modules-Social Distancing

Detects people in the monitored area and estimates the real-world distance between them, generating an alarm whenever two or more people are too close to each other for too much time. Distance and time thresholds can be configured to prevent false alarms or ignore short-term events, such as when people do not maintain a proper distance for just a few instants because their paths momentarily cross.

AiVu-Smart Modules-Speed

This module automatically estimates the speed of vehicles in transit through a virtual sensor. It allows to acquire data about traffic conditions along a road infrastructure (real-time estimation of the speed of each vehicle, minimum and maximum speed values detected by the sensor and average calculation).

AiVu-Smart Modules-Stop

This module generates an alarm signal each time stationary objects or vehicles are detected within a configured sensor area for longer than a defined time threshold. It can optionally ignore pedestrian. It detects vehicles or objects stationing on carriageways and/or inside tunnels, accidents, etc.

AiVu-Smart Modules-Cover

This rule detects camera coverings, caused for example by vandalism, dirty lenses or objects blocking the camera view, which might negatively affect image processing and interfere with video analytics.

AiVu-Smart Modules-Tamper

This rule detects camera movements, caused for example by tampering or vandalism, which might negatively affect image processing and interfere with video analytics.

AiVu-Smart Modules-Change Detection

This module, specifically designed for railway and subway scenarios, detects people and objects inside train carriages and can be employed to verify that carriages are empty before entering the depot. A one-of-a-kind algorithm, it’s ideal for driverless systems.

Download the AiVu-Smart Modules brochure!