Face Detection MTCNN Processor

Note	WORK IN PROGRESS

Options

face.detection.mtcnn.draw-face-annotations: (for `augmentation` outputMode only!) Draw face annotations on top of the inbound image. (Boolean, default: true)
face.detection.mtcnn.min-face-size: Minimal face size to detect in the input image. (Integer, default: 30)
face.detection.mtcnn.output-mode: Specifies the content type of the outbound message. (OutputMode, default: <none>, possible values: annotation,augmentation,alignment)
face.detection.mtcnn.scale-factor: (Internal config). Control the number of image pyramids used for the face detection (Double, default: 0.709)
face.detection.mtcnn.skip-face-annotation-header: (for `augmentation` outputMode only!) Skip the face annotations from to the outbound message header. (Boolean, default: false)
face.detection.mtcnn.step-one-threshold: (internal config). Threshold for the first stage of the detection (Double, default: 0.6)
face.detection.mtcnn.step-three-threshold: (internal config). Threshold for the third stage of the detection (Double, default: 0.7)
face.detection.mtcnn.step-two-threshold: (internal config). Threshold for the second stage of the detection (Double, default: 0.7)

Build involves two-stages. First build the apps and generate the binder specific app starters projects:

$ ./mvnw clean install -PgenerateApps

You can find the corresponding binder based projects in the apps subfolder. You can then cd into the apps folder:

$ cd apps

and build all binder projects

$ ./mvnw clean package

java -jar face-detection-mtcnn-processor.jar ... use the properties TODO

And here is a example pipeline that uses face-detection-mtcnn:

face-detection-mtcnn-stream= TODO