Processing Settings

app.domain.export.processing_settings

User-configurable settings for detection, tracking, and rendering pipelines.

`ProcessingSettings` `dataclass`

Mutable configuration snapshot for a video processing session.

ATTRIBUTE	DESCRIPTION
`detection_model_name`	YOLO model name, or `"None"` when disabled. TYPE: `str`
`min_detection_confidence`	Confidence threshold used to filter detections. TYPE: `float`
`chosen_labels`	Whitelist of class names to detect. TYPE: `list[str]`
`tracking_strategy`	Tracking algorithm name. TYPE: `str`
`tracking_source`	Input layer used for tracking. TYPE: `str`
`reid_embedding_dim`	Embedding size used by Re-ID tracking. TYPE: `int`
`min_iou`	IoU threshold for matching detections to tracks. TYPE: `float`
`min_tracker_confidence`	Confidence floor before pruning a track. TYPE: `float`
`confidence_decay`	Per-frame confidence decay for coasted tracks. TYPE: `float`
`offline_stitching_enabled`	Whether post-processing bridges fragmented tracks. TYPE: `bool`
`max_stitch_gap`	Maximum number of missed frames to bridge. TYPE: `int`
`gsi_smoothness`	Gaussian interpolation lengthscale. TYPE: `float`
`privacy_features_enabled`	Whether conservative privacy redaction rules are active. TYPE: `bool`
`resist_box_shrinking`	Whether track boxes resist sudden size contractions. TYPE: `bool`
`shrink_resistance`	Damping factor for shrinking track boxes. TYPE: `float`
`pre_detection_padding_sec`	Backfill duration before the first detection. TYPE: `float`
`post_detection_padding_sec`	Hold duration after the last detection. TYPE: `float`
`box_safety_padding_pct`	Proportional outward inflation of privacy boxes. TYPE: `float`
`draw_boxes`	Whether box outlines are shown in the editor preview. TYPE: `bool`
`box_border_width`	Preview outline width in pixels. TYPE: `int`
`show_preview_mask`	Whether to show the export blur mask in preview. TYPE: `bool`
`preview_blur_enabled`	Whether to render blur in the editor preview. TYPE: `bool`
`blur_enabled`	Whether Gaussian blur is applied during export. TYPE: `bool`
`blur_strength`	Blur sigma in pixels. TYPE: `float`

Note

Created per session with defaults, updated when the user changes UI controls, and converted to SessionSettingsViewModel through as_view_model() for widget binding.

Example

settings = ProcessingSettings( detection_model_name="yolov8n", min_detection_confidence=0.50, chosen_labels=["person", "car"], tracking_strategy="hungarian", blur_enabled=True ) view_model = settings.as_view_model() # For UI binding

Layer: Domain. Context: Represents application concepts without UI or infrastructure dependencies.

`blur_enabled: bool = False` `class-attribute` `instance-attribute`

`blur_strength: float = 15.0` `class-attribute` `instance-attribute`

`box_border_width: int = 2` `class-attribute` `instance-attribute`

`box_safety_padding_pct: float = 0.1` `class-attribute` `instance-attribute`

`chosen_labels: list[str] = field(default_factory=lambda: ['person', 'cat', 'dog'])` `class-attribute` `instance-attribute`

`confidence_decay: float = 0.05` `class-attribute` `instance-attribute`

`detection_model_name: str = 'None'` `class-attribute` `instance-attribute`

`draw_boxes: bool = True` `class-attribute` `instance-attribute`

`gsi_smoothness: float = 10.0` `class-attribute` `instance-attribute`

`max_stitch_gap: int = 25` `class-attribute` `instance-attribute`

`min_detection_confidence: float = 0.25` `class-attribute` `instance-attribute`

`min_iou: float = 0.3` `class-attribute` `instance-attribute`

`min_tracker_confidence: float = 0.1` `class-attribute` `instance-attribute`

`model_name_is_null: bool` `property`

Return whether no detection model is selected.

RETURNS	DESCRIPTION
`bool`	`True` when the model name is empty or the literal `"None"`.

`offline_stitching_enabled: bool = True` `class-attribute` `instance-attribute`

`post_detection_padding_sec: float = 0.5` `class-attribute` `instance-attribute`

`pre_detection_padding_sec: float = 0.5` `class-attribute` `instance-attribute`

`preview_blur_enabled: bool = False` `class-attribute` `instance-attribute`

`privacy_features_enabled: bool = False` `class-attribute` `instance-attribute`

`reid_embedding_dim: int = 256` `class-attribute` `instance-attribute`

`resist_box_shrinking: bool = True` `class-attribute` `instance-attribute`

`show_preview_mask: bool = False` `class-attribute` `instance-attribute`

`shrink_resistance: float = 0.85` `class-attribute` `instance-attribute`

`tracking_source: str = 'b'` `class-attribute` `instance-attribute`

`tracking_strategy: str = 'hungarian'` `class-attribute` `instance-attribute`

init(detection_model_name: str = 'None', min_detection_confidence: float = 0.25, chosen_labels: list[str] = (lambda: ['person', 'cat', 'dog'])(), tracking_strategy: str = 'hungarian', tracking_source: str = 'b', reid_embedding_dim: int = 256, min_iou: float = 0.3, min_tracker_confidence: float = 0.1, confidence_decay: float = 0.05, offline_stitching_enabled: bool = True, max_stitch_gap: int = 25, gsi_smoothness: float = 10.0, privacy_features_enabled: bool = False, resist_box_shrinking: bool = True, shrink_resistance: float = 0.85, pre_detection_padding_sec: float = 0.5, post_detection_padding_sec: float = 0.5, box_safety_padding_pct: float = 0.1, draw_boxes: bool = True, box_border_width: int = 2, show_preview_mask: bool = False, preview_blur_enabled: bool = False, blur_enabled: bool = False, blur_strength: float = 15.0) -> None

`as_view_model() -> SessionSettingsViewModel`

Create the widget-ready view of this session's processing settings.

RETURNS	DESCRIPTION
`SessionSettingsViewModel`	A view model with labels joined for direct text-field assignment.

Note

This conversion is a snapshot; later UI edits update the settings through the application service rather than mutating the snapshot.

Example

view_model = settings.as_view_model()

Processing Settings