Documentation for C-Werk 2.0.

Previous page Requirements for face detection tools  Fine-tuning the Face detection tool Next page

Attention!

You need to create a cache beforehand in order to start and correctly operate the basic Face detection tool on GPU (see Optimizing the operation of Face detection on GPU).

To configure the basic Face detection tool, do the following:

  1. Select the Face detection object. 

  2. If you need to use this detection tool for real-time face recognition, set the corresponding parameter to Yes (1, see Configuring real-time face recognition).

  3. If you need to record metadata, select Yes from the Record objects tracking list (2).
  4. If the camera supports multistreaming, select the stream for which detection is needed (3). For the correct operation of the Face detection, it is recommended to use a High-quality video stream.
  5. If you need to use this face detection tool in real-time together with FaceCube Recognition Server (see Configuring FaceCube integration), set Yes for the Real-time recognition on external service parameter (4).

  6. If you need to save age and gender information for each captured face in the database, select Yes in the corresponding field (1).

    Note

    The average error in age recognition is 5 years.

  7. If you use a bi-spherical XingYun lens, the detector will analyze two 180° spherical images by default (see Configuring fisheye cameras). This may decrease recognition quality. To dewarp the image before detection, select Yes for the Camera transform parameter (2). This parameter is relevant for other types of image transformation as well. 
  8. Select a processing resource for decoding video streams (3). When you select a GPU, a stand-alone graphics card takes priority (when decoding with NVIDIA NVDEC chips). If there is no appropriate GPU, the decoding will use the Intel Quick Sync Video technology. Otherwise, CPU resources will be used for decoding.
  9. Set the time (in milliseconds) between face search operations in a video frame in the Face detection period (msec) field (4). Acceptable values range is [1; 10000]. Increasing this value decreases the Server load, but can result in some faces being undetected. 
  10. If you plan to apply the masks detection tool, set Yes for the Face mask detection parameter (5, see Configuring Masks Detection).
  11. In some cases, the detection tool may mistake other object for a face. Select Yes in the Filter false alarms field (6) to filter out non-face objects,while calculating the vector model of a face and its recording into the metadata DB. If the filtering is on, false results will appear in the detection feed (see Face recognition and search), but will be ignored during searches in the archive.

  12. Analyzed framed are scaled down to a specified resolution (7, 1920 pixels on the longer side). This is how it works:

    1. If the longer side of the source image exceeds the value specified in the Frame size change field, it is divided by two.

    2. If the resulting resolution falls below the specified value, it is used further.

    3. If the resulting resolution still exceeds the specified limit, it is divided by two, etc.

      Note

      For example, the source image resolution is 2048*1536, and the specified value is set to 1000.

      In this case, the source resolution will be halved two times (512*384), as after the first division, the number of pixels on the longer side exceeds the limit (1024 > 1000).

      Note

      If detection is performed on a higher resolution stream and detection errors occur, it is recommended to reduce the compression.

  13. Specify the minimum and maximum sizes of the captured faces as a percentage of the frame size (8). 

  14. In the Minimum threshold of face authenticity field, set the minimum level of face recognition accuracy for the creation of a track (9). You can set any value by trial-and-error. No less than 90 is recommended. The higher the value, the fewer faces are detected, while the recognition accuracy increases.

  15. Select the processor for the face detection − CPU or NVIDIA GPU (10, see General information on configuring detection). 

    Attention!

    It may take several minutes to launch the algorithm on NVIDIA GPU after you apply the settings.

  16. If you use FaceCube integration (see Configuring FaceCube integration), activate the Send face images parameter (11).

  17. Enter the time in milliseconds after which the face track is considered to be lost in the Track loss time field (12). Acceptable values range is [1; 10000]. This parameter applies when a face moves in a frame and gets hidden behind an obstacle for some time. If this time is less than the set value, the face will be recognized as the same.

  18. If necessary, fine-tune the detection tool (see Fine-tuning the Face detection tool).

  19. In the preview window, set the rectangular area of the frame in which you want to perform face detection. To select the area, move the anchor points .

  20. Click the Apply button.

The basic Face detection tool is now configured.

  • No labels

Fine-tuning the Face detection tool

Optimizing the operation of Face detection on GPU