Aurora-AI

Artificial Intelligence (AI) Solutions using Deep Learning for automation and analysis

Company Profile

• Face Recognition

  We have built state-of-the-art face recognition that has processed millions of departing passengers at Heathrow, Manchester and more recently Senai and Hyderabad airports. This is based on capture of infrared images using our bespoke IR cameras which ensure that we always deliver high performance, high security solutions that operate in any of the conditions found at the airport.

  We also supply face recognition that works with visible light spectrum, so you can operate with any off-the-shelf camera to perform image capture. Whilst accuracy levels will always be lower than for our infrared systems, this provides an excellent solution where lighting levels are more consistent.

  We also supply face recognition that can compare stand colour photo (visible spectrum) images with those taken using infrared illumination. This provides the interoperability between different image acquisition technologies that airports demand.

  Summaries of current deployments can be viewed at:

  • AirAsia Fast Airport Clearance Experience System
  • British Airways’ Biometric gates at Heathrow

  We have also packaged our face recognition into products that support key process steps for the airport as follows.

  • Bio-Q for passenger queue monitoring.

  Bio-Q measures queue times between entry, way and exit points. People can be captured using cameras that view open areas or at control points. Very high penetration rates are achieved, especially when using our infrared cameras as they provide consistent results all the time, day or night, regardless of lighting. Faces are distributed to a central server whereby face matching takes place to generate time spans between specific locations. A key advantage of this Aurora product over rival systems is that it does not stream full video, removing the need for layout of extensive network infrastructure, greatly reducing cost of deployment.

  • Bio-Secure for identification at divestment.

  Bio-Secure provides a face recognition framework which enables integrators to identify passengers as they perform divestment at baggage inspection, based on enrolment images captured earlier in their journey. Identification messages are sent through a standard interface to enable the integrator to make decisions on the level of security checks that will be applied to the passenger and the trays that they used during the divestment process.

  • Bio-Frame for distributed biometric solutions.

  Bio-Frame is comprehensive software Framework and API, modular in design and extensible for the storage, distribution and matching of biometric templates. It can be configured to work in several operational topologies. It provides the back-end solution to enable any solution that relies on biometric template data for identification. Enables full compliance with the latest GDPR requirements.

  • Bio-Mobile for face recognition on mobile phones.

  Bio-Mobile provides face recognition for mobile platforms making use of onboard cameras to enable off airport enrolment ready for initial verification when the passenger arrives at the airport.

  For more information please see aurora-ai.com/ai-solutions/face-recognition/

  

• Predictive Analytics

  We apply Artificial Intelligence technology to a range of data analytics to produce outstanding accuracy. The features of this type of AI application include;

  • Prediction

  Proven solutions have been developed that use arrival and departure schedule data to predict the state of another variable in the future, for example, car park occupancy in 4 hours’ time. This is being extended to many other predictions to improve the Digital Twin of the airport and improve airport operations as a result.

  • Estimation

  Similarly to prediction, we are able to use AI regression techniques to generate estimated quantifications of live data. For example: the number of people present in an area; the gender and ethnicity ratios in a given population; or the remaining resource given the current situation.

  • Classification

  Our AIs are able to classify visual objects or data instances into categories according to highly complex feature patterns. As some examples, this technique has been applied to: the classification of guns, bombs and knives in x-ray imagery; recreational, VIP or fraudulent activity in online customer behavioural patterns; and field tagging in freeform PDF documents.

  More information can be found at aurora-ai.com/ai-solutions/data-analytics/

  

• Image Assessment

  We have developed AIs to enhance current processing of baggage.

  • Bag Tag OCR

  Our Bag Tag OCR AI reads text from images of luggage labels to enhance read rates during sortation and act as a failsafe when barcode reading systems fail. This is currently deployed in a major European hub airport to enhance remote processing of baggage.

  • Bag Classfication

  Through AI-based computer vision, classification and detection we have deployed systems that are able to automate luggage handling through classification as suitable for robotic loading to enable routing of bags that can be handled in this way to manual lines. Similar assessment to determine if a bag has straps and should be placed in a tub has also been assessed.

  We have also developed multiple AIs that use images to classify threat objects.

  • X-Ray Images

  Our AI can detect guns, knives and bombs based on the images used to train security officers. This is ready for extension based on images captured of real objects and would be easy to extend to cope with the new 3d images from the latest scanners.

  • Passive Emission

  We are working with a company that produces a new detector that can highlight threats on people with no requirement to remove coats or extract laptops from bags. Based on passive emission detection we have built AIs that can detect threats in real time as the subject moves past a detector.

  • Anomaly Detection

  AIs have been proven to learn what normal looks like for any data set and then to raise an alert when something different is processed. For example, this has been used to detect network threat intrusion on programmable logic controllers used for airport infrastructure management.

  More information can be found at http://aurora-ai.com/ai-solutions/threat-detection/ and http://aurora-ai.com/ai-solutions/bag-tag-ocr/

  

• Bespoke AI Development

  We have developed a simple engagement process that enables bespoke AI projects to be delivered quickly and at a very reasonable cost.

  We work with you to generate solutions that deliver business outcomes, enhance customer experience, and increase operational efficiency by improving the precision of predictions.

  More information can be found at http://aurora-ai.com/creating-an-ai-2/