kenzy_image · GitHub license Python Versions Read the Docs GitHub release (latest by date)

This module is dedicated to simplifying the interactions required for face detection, face recognition, object detection, and motion detection.


The easiest way to install kenzy_image is with the following:

pip install kenzy-image

Just make sure you're running Python 3.6 or newer.

Embedding into your program

Visit the detector page

Running as module

Options are as follows for starting kenzy_image:

python -m kenzy_image [OPTIONS]

General Options:
  -h, --help            show this help message and exit
  -c CONFIG, --config CONFIG
                        Configuration file
  -v, --version         Print Version

Startup Options:
  --camera-device       Device ID or RTSP stream to leverage for source images.
  --no-markup           Hide outlines and names
  --scale-factor SCALE_FACTOR
                        Image scale factor (decimal).  Values < 1 improve performance.
  --orientation ORIENTATION
                        Image orientation. (0, 90, 180, or 270)

Face Detection:
  --no-faces            Disable face detection
  --face-detect-default-name FACE_DETECT_DEFAULT_NAME
                        Set the Unknown face name
  --face-detect-model FACE_DETECT_MODEL
                        Model to leverage (hog or cnn)
  --face-detect-font-color FACE_DETECT_FONT_COLOR
                        Face names font color as tuple e.g. (0, 0, 255)
  --face-detect-outline-color FACE_DETECT_OUTLINE_COLOR
                        Faces outline color as tuple e.g. (0, 0, 255)
  --no-face-names       Hides the face names even if identified.
  --faces path name     Face image and name e.g. --face image.jpg LNXUSR1

Object Detection:
  --no-objects          Disable object detection
  --object-detect-type TYPE
                        Type of model to use (yolo or ssd)
  --object-detect-config OBJECT_DETECT_CONFIG
                        Object detection configuration
  --object-detect-model OBJECT_DETECT_MODEL
                        Object detection inference model file
  --object-detect-labels OBJECT_DETECT_LABELS
                        Object detection inference model label files
  --object-detect-font-color OBJECT_DETECT_FONT_COLOR
                        Object names font color as tuple e.g. (0, 0, 255)
  --object-detect-outline-color OBJECT_DETECT_OUTLINE_COLOR
                        Object detection outline color as tuple e.g. (0, 0, 255)
  --object-list  OBJECT_LIST
                        Limit list of objects to detect detection (optional)
  --no-object-names     Hides the object names even if identified.

Motion Detection:
  --no-motion           Disable motion detection
  --motion-detect-threshold MOTION_DETECT_THRESHOLD
                        Motion detection difference threshold
  --motion-detect-min-area MOTION_DETECT_MIN_AREA
                        Motion detection minimum pixel area
  --motion-detect-outline-color MOTION_DETECT_OUTLINE_COLOR
                        Motion area outline color as tuple e.g. (0, 0, 255)

Logging Options:
  --log-level LOG_LEVEL
                        Options are full, debug, info, warning, error, and critical
  --log-file LOG_FILE   Redirects all logging messages to the specified file

To start the services try:
python3 -m kenzy_image

More information available at:

The Object Detection model for ssd is MobileNet V3. The model for yolo is Yolov7 which will leverage any available and compatible cuda GPU. You may need to test both methods to find the one that suits your needs the best.

Help & Support

Help and additional details is available at