AIM-ES
Advanced Recognition Systems   >   Products    >   AIM-ES

Advanced Identity Matcher - Edge Source

An integrated multimodal matching solution for face and fingerprint recognition



AIM-ES

Fast - Accurate - Reliable

NEC's Advanced Identity Matcher - Edge Source (AIM-ES) uses the most accurate and NIST validated matching algorithms, our NeoFace® algorithm, to deliver an all-in-one solution capable of creating and administrating face and fingerprint recognition galleries and templates as well as providing identity matching and scoring.

Developed for small gallery sizes and easily deployable, AIM-ES is the fastest and most precise small-scale biometric engine available making it ideal for government agencies that require a highly dependable tool with industry-leading accuracy and search response times.

AIM-ES offers

  • Individual edge search nodes provisionable to hold a small set of enrolled images.
  • An interoperable Representational State Transfer (REST) application programming interface that permits communication via JavaScript Object Notation (JSON) as well as protocol buffers that can also be used to reduce overhead created by the serialization and deserialization of JSON objects.
  • Hardware agnostic capable of running on bare metal and in a cloud environment utilizing Docker, an independent software delivery platform that easily integrates into an existing environment and provides full stack portability for applications configurability.
  • A scalable response time capable of operating at 2 seconds or less on a gallery of up to 20,000 records.



    •  

    Dataset Management

    AIM-ES provides the robust functionality required to intuitively create and maintain memory-based biometric datasets.


    When a dataset needs to be created, a RESTful POST call is made to the dataset endpoint and a dataset ID is created.


    In-memory datasets can be populated with biometric templates either ad hoc by templatizing JPEG or PNG images or with pre-encoded biometric templates.


    All available datasets can be queried by using a RESTful GET call to the dataset endpoint with potential matching response times as fast as one second.


    Once the dataset is no longer needed it can be deleted with a DELETE call which frees up memory on the search node.

    Template Performance

    Template Creation
  • The calling application supplies an ID to identify the image, the image encoded in base 64, the minimum and maximum eye distance, minimum quality and maximum number of faces to find.
  • Templates for an image can be created by using a PUT call to the extract endpoint.
  • The extract endpoint can handle images with multiple faces and returns an array of facial features identifying each face in the photo by a bounding rectangle with face quality metrics.
  • Multiple photos can be supplied at one time and all will be encoded.
    •
    Template Extraction
  • Template extraction takes approximately 1.2 seconds per face.
  • The face features returned can then be stored or enrolled into a dataset on the order of milliseconds.
  • Templates can be created and stored in advance and quickly added to a dataset based on a list such as a passenger manifest.
  • Each template is approximately 4,500 bytes in size, therefore a gallery of 5,000 individuals will require approximately
  • Multiple templates per individual can be enrolled to increase accuracy.


    • Enrollment Functionality

    The use of preprogrammed command prompts makes the on-going maintenance of stored datasets easy to maintain and update.


    Update
    Enrollment is accomplished by using a POST call to the person�s endpoint of a dataset. Providing a dataset ID, either an image or facial feature set can be provided and enrolled.


    Delete
    An individual can also be removed from a dataset using a DELETE call in conjunction with the dataset ID and person ID.


    Retrieval
    For a list of existing enrolled individuals, a GET call with the dataset ID will return an array of enrolled subjects.

    Search Options


    Searching is accomplished by using a PUT call against the person's endpoint of a dataset, similar to an enrollment call. The search endpoint can accept a set of facial features or a raw image encoded in base 64.

    Searching a set of facial features against a database of facial features (1:N) can be accomplished in one second. Alternatively, a 1:1 search can be accomplished by providing the person ID within the dataset.

    The solution will return a response back to the requestor with photos, match and quality scores, and ranking of each identity.

     

    Proven Performance

    The National Institute of Standards and Technology (NIST) matching algorithm recognition benchmarks have consistently proven that NEC's biometric technologies have the fastest and most accurate face and fingerprint recognition algorithm and have the most resilient facial recognition technologies to viewing low angles, low resolution images and poor image quality.

    AIM-ES is a continuance in our commitment to providing cutting-edge solutions that deliver precise and trusted results through rapid identification and authentication of individuals.


    Questions?

    What advances can your organization achieve with AIM-ES?

    Let's Talk

    Questions?

    Looking To Finance This Solution
    NEC Financial Services can help with creative financing options tailored to your specific needs.
    Get Started

    Contact Us