Face Mask Evaluation

⚠️ Warning

This node is exclusive to the Forxai Mirror project and requires a specific installation for use.

Overview

The Face Mask Evaluation node analyzes face mask compliance by comparing model predictions against enabled check criteria. It identifies specific violations like missing masks or improper wearing (exposed nose or chin).

Use Face Mask Evaluation when:

• Monitoring public health guidelines in indoor or crowded spaces.
• Distinguishing between a total lack of a mask and improper wearing (e.g., "nose hanging out").

Inputs

Name	Type	Description
mask	prediction	Prediction for "mask worn correctly".
no_mask	prediction	Prediction for "no mask detected".
nose_exposure	prediction	Prediction for "nose exposed".
chin_exposure	prediction	Prediction for "chin exposed".
face_mask_enable	bool	Enables the check for missing masks.
nose_exposure_enable	bool	Enables the check for exposed noses.
chin_exposure_enable	bool	Enables the check for exposed chins.

Outputs

The outputs emitted depend on which "enable" inputs are set to true.

Name	Type	Description
face_mask_violation	bool	true if the dominant class is a violation (and enabled).
nose_exp_violation	bool	true if the nose is exposed (when enabled).
chin_exp_violation	bool	true if the chin is exposed (when enabled).

Parameters

This node does not have any configuration parameters. All behavior is controlled through inputs.

INFO

The node determines the dominant prediction by selecting the input with the highest probability (argmax). It does not use a numeric confidence threshold — whichever prediction class has the highest score wins. The enable inputs then control which violation outputs are emitted.

Example

Strict compliance monitoring

A building requires masks that cover both the nose and chin. All three enable inputs are set to true so the system checks every aspect of mask compliance.

Configuration:

• No parameters — all behavior is controlled via the boolean enable inputs.

Result: If the highest-probability prediction is nose_exposure, then face_mask_violation and nose_exp_violation are both true, while chin_exp_violation is false. If the winning class is mask, all violation outputs are false.

Related links

Reference:

• Face Orientation — analyzing head position and angles
• AI Model — basic model inference for detections