What is Machine Vision (sometimes called Vision AI or Computer Vision)?

There are many scenarios where physical, human inspection of assets is impractical, or where monitoring by video or camera produces data that cannot be usefully interpreted. CGI Machine Vision services and solutions work to change the economics of visual (physical or traditional sensor-based) inspection.

CGI Machine Vision can replace periodic manual inspection with continuous automatic analysis that can free up the staff performing monitoring to do more value-adding work and increase the rate of coverage by many times. In addition to monitoring fixed assets, CGI Machine Vision can also be applied to interpret vision data from traffic and crowd scenarios.

CGI Machine Vision provides a new source of business intelligence enabling a proactive approach to solving a wide range of operational challenges.

Drone flying in airspace

CGI’s Machine Vision solution is a combination of:

  • vision capture or other sensors (Internet of Things (IoT)) producing data,

  • at the point-of-capture ‘edge compute’ generative AI processing, and

  • analysis which draws on AI and Machine Learning to interpret photos and videos in a human-like way, in fact exceeding the capabilities of a ‘naked’ human eye
  • integration of the processed information into data operations.

CGI Machine Vision solutions can be applied in any industry, adding value to a wide range of scenarios. Anywhere that photos or videos are performing monitoring, CGI Machine Vision’s artificial intelligence vision analysis can help.

 

Contact a CGI Machine Vision expert

A CGI Machine Vision solution can be tailored to many different scenarios. Our CGI Machine Vision platform can be deployed on a range of embedded Internet of Things (IoT) devices where image data is being captured. We have experience extracting information from a range of sources including:

  • Edge compute board with attached camera
  • CCTV
  • Cameras and surveillance equipment
  • Drones
  • Thermal Cameras
  • Ultrasonic sensors
  • High speed cameras
  • Night vision recording devices

 

 

 

Machine Vision Graphical Processing Unit

The images are processed by a Graphical Processing Unit (GPU) that is at the site where the recording is being captured (‘edge computing’, or edge AI) using Artificial Intelligence – generative and deep learning models. CGI can supply GPUs as part of your solution, including, where appropriate, solar panels to provide an off-the-grid or carbon neutral solution.

The encrypted processed data is then sent via our secure cloud hub. The CGI Machine Vision edge-compute processing solves the issues of data transmission delay and bandwidth concerns. Only the relevant, processed information is pushed to data operations, providing you with real-time data analysis tailored to your project objectives. 

We can network many devices together to monitor all kinds of operations, from wastewater treatment plants and smart water meters to supporting entire Smart Cities or remote infrastructure. By combining these devices, we can ensure that all key infrastructure (for example sewers, bin collections, and other citizen services) run efficiently and effectively.

For businesses leveraging CGI Machine Vision’s artificial intelligence analysis, real-time responses become possible, enabling the adoption of predictive and proactive operational models. Asset resilience and availability is enhanced, and sustainability, safety, and regulatory compliance are also significantly improved.

Camera monitoring remote infrastructure

Machine Vision can:

  • Increase data quality by improving the velocity, accuracy, efficacy, and efficiency of visual inspection
  • Increase the availability and speed of real-time data analysis
  • Reduce latency of sending image and video streams to the cloud
  • Reduce cloud storage and data transfer costs incurred sending data to the cloud for analysis
  • Capture more data points, more frequently, without increasing the cost of data
  • Deploy always-on, automated solutions that are not dependent on human resources
  • Implement use cases, previously impractical and/or impossible

CGI Machine Vision can be applied anywhere where vision data is being captured. Here are some scenarios we are currently working with our clients on where the proactive alarming provided by the CGI Machine Vision solution is reducing operational costs, ensuring regulatory compliance and improving safety.

People Dynamics

People Dynamics
  • Counting people at events and hosting facilities
  • Monitoring crowds at stations to keep public transport passengers safe
  • Retail interest ‘heat maps’ to inform product and services placements

Infrastructure

Infrastructure

  • 24x7 Monitoring of unmanned facilities such as water treatment plants, dams, pipes and substations to produce alerts for stopped or slow equipment, water leaks, or to produce alerts for unusual activity e.g. people.

Flood & Erosion Monitoring

Flood & Erosion Monitoring
  • Monitoring the integrity of remote or physically difficult to access assets such as bridges and railways.
  • Producing alerts about property damage due to for example, flooding or vegetation encroachment or wear and tear.

Manufacturing and Agriculture

 

Manufacturing and Agriculture

  • Real-time analysis of production line and item quality
  • Crop monitoring
  • Autonomous monitoring of multiple operations simultaneously
  • Gauging production volumes and outcomes

Smart Cities

 

Smart Cities

  • Dynamic asset discovery
  • Alerts for vandalism and graffiti
  • Monitoring garbage and bin collections
  • Parks and roads asset monitoring 

Traffic

 

Traffic

  • Detect impending bridge strike, or level crossing issues before vehicle impact
  • Extraction of vehicle registration plate details in real time
  • Gauge volume and loads of vehicles
  • Detect loose or unsecured loads
  • Supports multiple lanes of traffic

Bird and Wildlife Watch

 

Bird and Wildlife Watch

  • Detecting bird and flock numbers
  • Environmental threat analysis
  • Detect birds, bats and other wildlife that can cause damage to an aircraft
  • Provide alerts for remotely monitored traps, burrows, nests, hives

Smart Meters

 

Smart water meter

  • Near real-time water usage measurements
  • Intelligent tracking of water usage source e.g. showers, toilet
  • Warning of leaks and other disruptions
  • Digital readers can be added to existing meters without need for plumbers

Construction

 

Construction

  • Improve worker safety and security compliance through people video analytics:

    • monitor use of safety and protective clothing
    • detect workplace accidents
    • spot potential hazards ahead of time
  • Vehicle site visit tracking

 

What is Edge Computing?

Edge computing is a distributed computing paradigm that brings computational processing and data storage closer to the location where it is needed, which is typically at or near the "edge" of a network. In contrast to traditional cloud computing, where data and processing are centralized in remote data centres, edge computing pushes these capabilities to the periphery of the network, closer to the data source or end-user device.

What are the advantages of keeping the processing close to the point of capture (Edge Computing)?

Deploying local Graphical Processing Units (GPU) results in:

  • Low Latency: Edge computing reduces the latency or delay in data transmission because data doesn't have to travel to a distant data center and back. This is crucial for applications that require real-time or near-real-time processing, such as autonomous vehicles, industrial automation, and augmented reality.
  • Bandwidth Efficiency: By processing data locally at the edge, only relevant or processed information needs to be sent to the cloud, reducing the amount of data transmitted over the network. This can lead to more efficient use of network resources.
  • Data Privacy and Security: Edge computing can enhance data privacy and security by keeping sensitive data closer to its source and reducing the risk of data breaches during transit. Data can be processed and anonymised locally before sending it to the cloud.
  • Scalability: Edge computing can be highly scalable because additional edge devices can be added to distribute the processing load, making it suitable for applications with variable or unpredictable workloads.
  • Reliability: In edge computing, individual edge devices can continue to operate even if they lose connection to the central cloud or data center. This improves the reliability and availability of applications in remote or challenging environments.
  • Offline Operation: Edge devices can operate autonomously without a constant internet connection, allowing them to function in scenarios with intermittent connectivity or limited network access.
  • Real-Time Decision Making: Applications that require real-time decision-making, such as monitoring and controlling industrial machinery, benefit from edge computing because they can make decisions locally without relying on cloud-based processing.
How does CGI Machine Vision work with Digital Twins?

A digital twin is a virtual representation of a physical object, system, or process. It is a digital counterpart that mirrors the physical entity in a digital form, providing a real-time or near-real-time reflection of its status, behavior, and performance.

CGI Machine Vision uses a digital twin first philosophy. Data is captured and then securely and efficiently synchronized, from the point of capture to a live digital model.

What cyber security provisions have been built in?

Security is core to Machine Vision including Secure Boot on devices, signed boot files, encrypted file systems and integrated OOTA updates.

Machine Vision has been designed with the principles of Secure-by-Design and Secure-by-Default.

How does Generative AI add value to computer vision?
  • Anomaly Detection: Generative models can be used for anomaly detection in video data by modeling the normal behavior and flagging deviations from this baseline. This is valuable in surveillance, security, and industrial applications.
  • Video Enhancement: Generative models can enhance the quality of videos by reducing noise, improving resolution, and enhancing details. This is useful for tasks like upscaling, denoising, and restoring old or degraded video footage.
  • Data Augmentation: Generative models can generate synthetic data, which can be used to augment training datasets. This helps improve the robustness and generalization of computer vision models, especially when labeled data is scarce or costly to obtain.
  • Privacy Preservation: Generative models can generate synthetic data that closely resembles real data while protecting individual privacy. This is particularly important when handling sensitive image data.
Who are CGI’s Australian Machine Vision partners?

Our Machine Vision technology partners include:

  • Microsoft Azure Digital Twin
  • Nvidia