SentiVeillance Cluster

Ready-to-use software for easy integration of biometric face identification, vehicle and pedestrian classification and tracking, as well as automatic license plate recognition.

Features

Real-time performance

The system executes face recognition, pedestrian or vehicle classification, and tracking in real time, even when an input video is used instead of a live stream. All video streams are being analyzed in parallel.

Cluster architecture

Several machines can be connected to a cluster to process a larger number of video streams. The solution allows integration into surveillance systems with any number of surveillance cameras and supports multiple GPUs.

Search operations

The software permits sending search requests in the events log to the linked SentiVeillance Cluster via the VMS user interface, which returns the search results.

Multiple modalities for surveillance systems

Biometric face recognition, pedestrian/vehicle detection and tracking (VH), and automated license plate recognition (ALPR) modalities.

Technical details

Face size

4 % of the frame’s larger side (at least 32 pixels) is the minimal recommended distance between eyes for a face on video stream. The performance depend on actual size of a face in a video stream, not on the size of the whole frame.
Face record size in the watch list is less than 0.5 kilobytes. A person can have multiple face records stored.
The matching against a watch list is done in less than 0.5 seconds when the watch list has 20,000 face records.

Face posture

SentiVeillance Cluster assures accurate face detection and tracking when the face posture meets these constraints:

  • head roll (tilt) – ±15 degrees from frontal position;
  • head pitch (nod) – ±25 degrees from frontal position (several face views should be enrolled);
  • head yaw (bobble) – ±45 degrees from frontal position (several face views should be enrolled).

Face enrollment

  • Image quality during enrollment is important, as it influences the quality of the face template.
  • Several images during enrollment are recommended for better facial template quality which results in improvement of recognition accuracy and reliability.
  • Additional enrollments may be needed when facial hair style changes, especially when beard or mustache is grown or shaved off.

System requirements

System requirements for video management systems (VMS)

SentiVeillance Cluster software uses face biometrics to perform persons identification and tracking in video streams, which are received from VMS. The results of SentiVeillance Cluster operations are sent back to the VMS.
SentiVeillance Cluster is designed to be used with the following video management systems (VMS):

  • Milestone XProtect VMS. The plug-in for this VMS can be customized, as its source code is provided.
  • Standalone Generic VMS client (sample sources are provided) can also be used on Microsoft Windows platform. This client accepts video streams via RTSP (Real Time Streaming Protocol) with some specific requirements:
    • Only RTP over UDP is supported.
    • H.264/MPEG-4 AVC or Motion JPEG should be used for encoding the video stream. 

A VMS should be deployed on hardware which provides enough resources for generating specified resolution video streams and storing video for a specified time frame. Please refer to VMS manufacturers for specific system requrements.

System requirements for SentiVeillance Cluster software

There are the following system requirements for each machine in a SentiVeillance Cluster based system:

  • PC or server hardware with x86-64 (64-bit) compatible processor:
    • 3 GHz or better processor with 6 cores is recommended.
    • AVX2 support is required. Most modern processors support this instruction set, but please check if a particular processor model supports it.
  • A graphical processing unit (GPU) is required.
    • NVIDIA GeForce RTX 3080 GPU or better is recommended for systems with up to 10 cameras.
    • at least 6 GB of VRAM is recommended.
      Compute Capability 3.5 or better should be supported by the GPU.
    • CUDA 11.x toolkit or newer is required
    • cuDNN 7.5 library is required.
  • 32 GB of RAM is recommended.
  • Network connection. At least 1 gigabit per second data transfer rate is recommended for real-time processing of multiple high resolution video streams.
  • Linux specific:
    • Debian 11.2 OS
    • glibc 2.17 or newer
    • GStreamer 1.10.x or newer with gst-vaapi plugins installed for hardware accelerated video decoding
    • libgudev-1.0 219 or newer
    • wxWidgets 3.0.0 or newer libs

Related products

VeriLook facial identification technology is designed for biometric systems developers and integrators. The technology assures system performance and reliability with live face detection, simultaneous multiple face recognition and fast face matching in 1-to-1 and 1-to-many modes.

MegaMatcher Accelerator is a hardware/software solution that provides high speed, high volume biometric identification for national-scale projects.

Neurotechnology Face Verification system is designed for integration of facial authentication into enterprise and consumer applications for mobile devices and PCs.

A high-performance biometric identification solution developed in-house by Neurotechnology using award-winning technologies. It is ready for immediate use or can be tailored to meet your specific business needs and includes all necessary components for government and enterprise applications at worldwide scale.

Download 30-day trial

A free trial is available to download, and the long-term solution can be acquired online or through our distributor.