facedetect

Performs face detection on videos and images. If you have high cpu load you need to use videoscale with capsfilter and reduce the video resolution.

The image is scaled down multiple times using the GstFaceDetect::scale-factor until the size is <= GstFaceDetect::min-size-width or GstFaceDetect::min-size-height.

Example launch line

 gst-launch-1.0 autovideosrc ! decodebin ! colorspace ! facedetect ! videoconvert ! xvimagesink

Detect and show faces

 gst-launch-1.0 autovideosrc ! video/x-raw,width=320,height=240 ! videoconvert ! facedetect min-size-width=60 min-size-height=60 ! colorspace ! xvimagesink

Detect large faces on a smaller image

Hierarchy

GObject
    ╰──GInitiallyUnowned
        ╰──GstObject
            ╰──GstElement
                ╰──GstBaseTransform
                    ╰──GstVideoFilter
                        ╰──GstOpencvVideoFilter
                            ╰──facedetect

Factory details

Authors: – Michael Sheldon

Classification:Filter/Effect/Video

Rank – none

Plugin – opencv

Package – GStreamer Bad Plug-ins

Pad Templates

sink

video/x-raw:
         format: RGB
          width: [ 1, 2147483647 ]
         height: [ 1, 2147483647 ]
      framerate: [ 0/1, 2147483647/1 ]

Presencealways

Directionsink

Object typeGstPad


src

video/x-raw:
         format: RGB
          width: [ 1, 2147483647 ]
         height: [ 1, 2147483647 ]
      framerate: [ 0/1, 2147483647/1 ]

Presencealways

Directionsrc

Object typeGstPad


Properties

display

“display” gboolean

Sets whether the detected faces should be highlighted in the output

Flags : Read / Write

Default value : true


eyes-profile

“eyes-profile” gchararray

Location of Haar cascade file to use for eye-pair detection

Flags : Read / Write

Default value : /usr/share/opencv4/haarcascades/haarcascade_mcs_eyepair_small.xml


flags

“flags” Opencv-face-detect-flags *

Flags to cvHaarDetectObjects

Flags : Read / Write

Default value : do-canny-pruning


min-neighbors

“min-neighbors” gint

Minimum number (minus 1) of neighbor rectangles that makes up an object

Flags : Read / Write

Default value : 3


min-size-height

“min-size-height” gint

Minimum area height to be recognized as a face

Flags : Read / Write

Default value : 30


min-size-width

“min-size-width” gint

Minimum area width to be recognized as a face

Flags : Read / Write

Default value : 30


min-stddev

“min-stddev” gint

Minimum image average standard deviation: on images with standard deviation lesser than this value facedetection will not be performed. Setting this property help to save cpu and reduce false positives not performing face detection on images with little changes

Flags : Read / Write

Default value : 0


mouth-profile

“mouth-profile” gchararray

Location of Haar cascade file to use for mouth detection

Flags : Read / Write

Default value : /usr/share/opencv4/haarcascades/haarcascade_mcs_mouth.xml


nose-profile

“nose-profile” gchararray

Location of Haar cascade file to use for nose detection

Flags : Read / Write

Default value : /usr/share/opencv4/haarcascades/haarcascade_mcs_nose.xml


profile

“profile” gchararray

Location of Haar cascade file to use for face detection

Flags : Read / Write

Default value : /usr/share/opencv4/haarcascades/haarcascade_frontalface_default.xml


scale-factor

“scale-factor” gdouble

Factor by which the frame is scaled after each object scan

Flags : Read / Write

Default value : 1.25


updates

“updates” Face-detect-updates *

When send update bus messages, if at all

Flags : Read / Write

Default value : every_frame (0)


Named constants

Face-detect-updates

Bus messages update scheme

Members

every_frame (0) – Send update messages on every frame
on_change (1) – Send messages when a new face is detected or one is not anymore detected
on_face (2) – Send messages whenever a face is detected
none (3) – Send no messages update

Opencv-face-detect-flags

Flags parameter to OpenCV's cvHaarDetectObjects function.

Members

do-canny-pruning (0x00000001) – Do Canny edge detection to discard some regions

The results of the search are