MXSERVER, from Tygart Technology,  is a Big Data cloud-enabled and virtualised video and photographic media analysis solution incorporating facial recognition.

• It can be run and maintained in most virtualised environments and can be hosted in a public cloud, private cloud or installed on local servers.
• It is completely browser-based, allowing multi-user role-based authentication with multiple public or private watchlists.
• It has a complete set of programming interfaces to allow integration into other systems and workflows allowing it to be run as a “black-box” engine.
• Using MXSERVER, all faces from the ingested video or photographs are:
  • Extracted and cropped.
  • Searched against a watchlist.
    (Matches resulting in watchlist hits which can be pushed to external systems)
  • Indexed so the media can be searched with a photograph.
    (where have we seen this person before?)
• Media inputs to MXSERVER may include:
  • Live surveillance cameras (CCTV).
  • Body Worn Video.
  • Archived video.
  • Online Sources (YouTube, Facebook etc).
  • Confiscated hardrives, phones, PCs (Digital Forensics).
  • Live from police mobile phones.
  • Members of the public.
• MXSERVER is linearly scalable, simply by adding / removing “worker” virtual machines.
• Allevate also offers a low-bandwidth live surveillance option that moves some processing next to the camera to:
  • Perform face detection, pose correction, best face selection.
  • Create JPGs of individually cropped faces.
    Therefore, only JPG images of faces seen in the video stream are sent over the network to MXSERVER.
  • https://allevate.com/facesearcher/
• MXSERVER is independent of the face recognition algorithm and is already integrated with face recognition algorithms of multiple vendors.

MXSERVER™ is a virtualised video / photographic media analysis solution incorporating AI that leverages cloud-technology and deep learning-based face detection and recognition algorithms to rapidly analyse and search massive collections (>100M) of videos/photos.

Combined with Edge detection, it can monitor large arrays of video surveillance cameras in real time to deliver “Ubiquitous identification”, using a single centralised matching infrastructure.

MXSERVER’s architecture simplifies development, integration, operations and maintenance efforts by abstracting the complexity of a flexible and massively scalable computing architecture. It enables accelerated development timelines and significantly reduced costs when compared to custom or algorithm dependent solutions. Leveraging a COTS service-oriented cloud-based computing platform simplifies and dramatically shortens the time (within hours/days) needed to install, configure and begin operating a high-performance computing environment.

MXSERVER can be run in most virtualised environments and can be hosted in a public cloud, private cloud or installed on local servers. It is completely browser-based, allowing multi-user role-based authentication with multiple public or private watchlists.

It is algorithm agnostic, leveraging best-of-breed detection and matching algorithms. Over time, no single vendor will be able to effectively serve as the best face, fingerprint and iris algorithm provider.

This vendor agnostic approach utilises the matcher algorithm as a plug-and-play commodity, enabling a flexible biometric search service that can be quickly upgraded to deliver better performing algorithms when available. Simply replace the algorithm, not the entire biometric search core – avoiding a costly re-engineering of the biometric system.

MXSERVER implements a microservice architecture leveraging Docker containers and Swarm to deliver a video/photo analytic and biometric search capability with massive processing capacity and speed. Containerising services facilitates the rapid integration of new video analytic services and supports plug- and-play of improved algorithms – enabling MXSERVER to deliver processing pipelines that are scalable, flexible and run services/algorithms at unprecedented speeds.