The requirements for the cameras that will work with the Neurotracker module are listed in the following table.

Camera

  • Resolution: 640х480 pixels.
  • Correct operation is not guaranteed with fish-eye lenses
  • Fps: not less than 2 (recommended: 4 or more)

Lighting

  • The objects must be visually separated from the background and from each other

Scene and camera angle

  • The camera shall be installed at an angle at a height of at least 3 meters, or on the ceiling, or facing strictly down (for counting visitors)

Objects image

  • Minimal object size: at least 10% of the frame area.

Hardware platform requirements:

  1. NVidia and Intel® HD Graphics GPU are supported.
    CPU resources and neural network accelerators based on Intel Myriad X processors, for example, Intel NCS and Intel HDDL, can be used.
  2. If CPU or Intel GPU is selected for analytics operation, take into account that the following processors are supported: Intel Core and Intel Xeon 6-8 generation and Intel Pentium® N4200/5, N3350/5, or N3450/5 with Intel® HD Graphics https://software.intel.com/en-us/openvino-toolkit/hardware.
  3. NVIDIA GeForce 1050 Ti GPU or newer. GPU requirements:
    1. at least 2 GB of memory;

    2. Compute Capability 3.0 or higher.

      Check the Compute Capability version for the video card on the manufacturer's web site.


One neural network uses 500 MB of video memory when using a video card. For example: to operate any number of channels of a fire detection tool based on neural networks and any number of channels of a smoke detection tool based on neural networks, use a video card with 1GB of memory or more. Several video cards can be installed in one system.