Video analytics systems
Development of algorithms and software for the intellectual analysis of digital images in the field of video surveillance and security: recognition, detection, visualization of objects or panoramas, analysis and image search, etc.
Related technologies
- Machine learning
- Internet of things
Scope of applicability
Video Surveillance and SecurityTasks
- Analysis of a large number of objects
- Continuous operation without loss of quality
- High speed and accuracy of measurements
- Automatic image analysis
- Automatic decision-making
- Performance of additional functions as the task may require
Advantages
- Improving the speed and quality of control by eliminating the “human factor”
- multitask capability (detection, classification, automated decision-making, matching search, etc.)
- Reduction of labor costs and other costs of traditional verification methods
- Guaranteed detection of emergencies and violators; the database search option.
Our project approach
We develop algorithms and software for processing video streams and images to solve a wide range of tasks, depending on the application area and the customer’s goals. A separate area is the development and optimization of software solutions for embedded systems with limited computing capacities.Technical vision systems
Intelligent analysis of digital images from industrial cameras to improve the accuracy and efficiency of equipment and processes, often used as a component of larger systems and focused on specific application tasks.
Related technologies
- Machine learning
- Industrial Internet of Things
Scope of applicability
- Video Surveillance
- Industrial automation
- Robotics
- Vehicle control systems
- Medical information systems
Tasks
- Analysis of a large number of objects
- Continuous operation without loss of quality
- High speed
- High measurement accuracy
- Automation of verification and control
Advantages
- Improving the speed and quality of control
- Stable quality due to the exclusion of the “human factor”
- Multi-task capability (detection, classification, identification of objects, automated decision-making, etc.)
- Reduction of labor costs and other costs on traditional verification methods
- Optimization of business processes and guarantee of their effectiveness
Our project approach
We carry out a full cycle of hardware and software development; develop and adapt machine vision algorithms to the performance targets. Orientation to the target software and hardware platform improves the performance of algorithms. In the development, we use machine- learning and deep-learning approaches.Multisensory systems, IIoT
We develop systems for collecting, transmitting and analyzing data from transmitters and sensors for integrated control of equipment and processes. As “input devices” of primary information, video cameras, gyro sensors, barometers, sound, temperature, and pressure sensors, etc. are used. “Dark” data are conveyed to a cloud computing service for interpretation and analysis. The result is visualized on the cameraman's workstation in the form of formalized data on the status of key processes.
Related technologies
- Machine learning
- Internet of things
Scope of applicability
It is used in production, in the operation of critically important engineering structures (pipelines and gas pipes) in difficult for access conditions, wherever comprehensive monitoring of equipment or processes is required.
Tasks
- Collection and processing of a large amount of heterogeneous data
- “Instant cut” of information simultaneously throughout the system
- Continuous monitoring of key parameters in the real-time mode
- The possibility to optimize system management using additional functions of analytics including the predictive one.
Advantages
- Enhancement of the quality and effectiveness of monitoring of the state of objects
- Ability to control the operation process using forecasts and recommendations
- Capacity to prevent critical situations before they take place
- Reduction of time and financial costs for traditional control methods
- Reducing the costs of operating and repairing the system
Our project approach
- Solving complex engineering problems in the development of data-acquisition units with special requirements for the performance, accuracy, cost, energy consumption, etc.
- Development of intelligent embedded software for preprocessing, data compression, and optimization of equipment operating modes
- Enhancement of the speed of data collection and transmission processes based on boundary computing: data processing is performed not in the data center, but directly on the IIoT monitoring devices.
Laser scanning data processing
Laser scanning is a modern and rapidly developing type of remote sensing, characterized by high accuracy and density of measurements. The software for processing scanning results, i.e., point clouds, should increase productivity; it also has the functions of analyzing three-dimensional spatial information according to specified parameters depending on the customer’s goal.
Related technologies
- Information modeling of buildings and structures
Scope of applicability
Construction, design, manufacturing, electric power, environmental monitoring, surveying, development of applications with augmented reality.Tasks
- Obtaining information about the object with the possibility of deep detailing
- Creating virtual elevation models and objects
- Image analysis by necessary parameters
- Surveying hazardous and inaccessible objects from a safe distance
- Visualization of the results
Advantages
- Information completeness: millions of points with three-dimensional coordinates
- Ability to automate the shooting
- Reduction of time and financial costs on performing high-precision measurements
- The possibility of repeated checks.
Our project approach
We develop algorithms and software tools for hybrid processing of large amounts of data in the real- time mode. The web interface enables remote data processing. Hybridization consists in combining and jointly processing data from different sources. Clouds of points from mobile, ground or air laser scanners and raster images obtained from cameras, when superimposed, allow selecting the geometry of objects.Machine learning technologies
A system for collecting and processing data on the state of equipment or processes based on mathematical models and algorithms that allow to build predictive analytics based on historical data on the system functioning.
Related technologies
- Technical vision systems
- Video analytics systems
- Internet of things
Scope of applicability
Banking, IT, retail, industry, logistics, agriculture and crop production.Tasks
- Monitoring the status of equipment and processes
- immediate interpretation of large amounts of data
- Decision-making support
- Forecast of emergencies, reduction of risks
- Assistance in solving applied problems
Advantages
- Improving the efficiency of parameter monitoring
- Reduction of the decision-making time
- Reduction of the number of emergencies due to their forecasting
- Reduction of the equipment repair and maintenance costs
- Identification of bottlenecks and optimization of business processes for the increase of the rate and efficiency of the company’s operation
- Development focused on the application tasks of a particular company
Our project approach
We use a wide range of machine learning algorithms to achieve customer goals: from decision trees to ultra-precise neural networks. We have rich experience in the work on the analysis of the video stream using machine learning algorithms. If you need to implement solutions on embedded platforms, we transfer resource-intensive solutions to platforms with limited computing capacities.
Neurointerfaces
Neurocomputer interfaces are the devices for the exchange of information between the brain and an external device. The technology is based on recording the electrical activity of the brain and converting it into commands for external devices.
Related technologies
- Internet of things
- Vision Systems
Scope of applicability
Medicine, robotics, games and virtual reality devices, video surveillance, security control systems, IT
Tasks
- Reading brain signals, decoding and transmitting signals of the electrical activity of the brain to the device
- Control of external devices using brain impulses
- Control of brain activity using impulses from external devices (invasive interfaces).
Advantages
- Rehabilitation and treatment of patients with CNS injuries
- Strengthening intelligence, increasing brain activity, improving learning capacity
- Reducing the time of reaction
- Ability of remote control of things
- Possibilities for integration into security systems and decision-making
Our project approach
The field of neural interfaces development is quite old and, at the same time, relatively young. Such a contradiction is acceptable due to the fact that while the mankind has learned how to relatively stably receive the initial signal yet in the 20th century, the methods for its analysis are actively developing these days.
In contrast to invasive sensors used in the medical bioengineering for direct implanting into the brain, we are developing the line of non-invasive neural interfaces that record the electrical activity of the brain from the head surface; those require accurate algorithms for successful processing of the recorded information.
Our competencies allow us to create devices that record the initial signal, to implement methods for its filtering, pre-processing and analysis.