haybale-visiongraph 0.0.26__tar.gz

haybale_visiongraph-0.0.26/.gitignore

@@ -0,0 +1,28 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# Distribution / packaging
+build/
+dist/
+*.egg-info/
+*.egg
+
+# Virtual environments
+.venv/
+venv/
+env/
+
+# Test / type-check caches
+.pytest_cache/
+.ruff_cache/
+.mypy_cache/
+.tox/
+.coverage
+htmlcov/
+
+# Editor / OS
+.DS_Store
+.idea/
+.vscode/

haybale_visiongraph-0.0.26/PKG-INFO

@@ -0,0 +1,65 @@
+Metadata-Version: 2.4
+Name: haybale-visiongraph
+Version: 0.0.26
+Summary: Visiongraph library for Haywire node system demonstrating custom types, nodes, widgets, and renderers
+Author: Haywire Team
+License: MIT
+Keywords: haywire,node-editor,visiongraph,visual-programming
+Requires-Python: >=3.12
+Requires-Dist: depthai~=2.30
+Requires-Dist: fastapi
+Requires-Dist: haybale-core>=0.0.32
+Requires-Dist: haywire-core>=0.0.32
+Requires-Dist: nicegui>=3.12.1
+Requires-Dist: numpy
+Requires-Dist: opencv-contrib-python>=5.0.0.93
+Requires-Dist: typing-extensions>=4.15.0
+Requires-Dist: visiongraph[mediapipe,onnx,openvino]
+Description-Content-Type: text/markdown
+
+# Visiongraph
+
+<!-- marketstall:share-url:start -->
+```sh
+# Released packages (recommended):
+https://going-haywire.github.io/haybale-visiongraph/marketplace.toml
+
+# Always the latest (tracks the current branch):
+https://github.com/going-haywire/haybale-visiongraph/blob/master/marketstall.toml
+
+# Frozen to this version:
+https://github.com/going-haywire/haybale-visiongraph/blob/v0.0.26/marketstall.toml
+```
+<!-- marketstall:share-url:end -->
+
+Visiongraph library
+
+## Nodes
+### Vision
+- **Annotate Results** — Draw estimator results (boxes / masks / poses) onto a frame
+- **Frame Event** — Triggered when a camera frame is ready; exposes colour/depth/infrared streams
+- **Frame Info Display** — Displays information about frames with live preview
+- **OAK-D Camera** — Opens an OAK-D depth camera and emits colour/depth/infrared frame callbacks
+- **Object Detector** — Detect objects in a frame (bounding box + class + score)
+- **Pose Estimator** — Estimate human body pose (named joints + skeleton) per person
+- **Segmentation** — Instance segmentation: detect objects and their pixel masks
+- **Tracker** — Assign stable tracking ids to detections across frames
+- **Web Camera** — Starts a webcam stream and emits frame callbacks
+
+## Types
+- **Depth Frame** — Single-channel uint16 metric depth buffer (millimetres)
+- **Detection Result** — Object detections: bounding box, class, score (and tracking id)
+- **Gray Frame** — Single-channel uint8 luminance video frame (e.g. infrared)
+- **Landmark Result** — Landmark detections: a detection plus a set of landmark points
+- **Multiframe Callback** — Subscription for multi-frame streams
+- **Pose Result** — Human pose: landmarks with named joints and skeleton connections
+- **RGB Frame** — 3-channel uint8 colour video frame
+- **Segmentation Result** — Instance segmentation: detection plus a per-instance mask
+- **Vision Result** — A list of estimator results (base type for all result kinds)
+
+## Widgets
+- **NumpyViewerWidget** — Streaming video viewer for numpy arrays using custom StreamingViewer
+
+## Adapters
+- **GrayToRgbAdapter** — Replicate a single-channel grey frame to a 3-channel colour frame
+- **RgbToGrayAdapter** — Convert a 3-channel colour frame to a single-channel grey frame (luminance)

haybale_visiongraph-0.0.26/README.md

@@ -0,0 +1,46 @@
+# Visiongraph
+
+<!-- marketstall:share-url:start -->
+```sh
+# Released packages (recommended):
+https://going-haywire.github.io/haybale-visiongraph/marketplace.toml
+
+# Always the latest (tracks the current branch):
+https://github.com/going-haywire/haybale-visiongraph/blob/master/marketstall.toml
+
+# Frozen to this version:
+https://github.com/going-haywire/haybale-visiongraph/blob/v0.0.26/marketstall.toml
+```
+<!-- marketstall:share-url:end -->
+
+Visiongraph library
+
+## Nodes
+### Vision
+- **Annotate Results** — Draw estimator results (boxes / masks / poses) onto a frame
+- **Frame Event** — Triggered when a camera frame is ready; exposes colour/depth/infrared streams
+- **Frame Info Display** — Displays information about frames with live preview
+- **OAK-D Camera** — Opens an OAK-D depth camera and emits colour/depth/infrared frame callbacks
+- **Object Detector** — Detect objects in a frame (bounding box + class + score)
+- **Pose Estimator** — Estimate human body pose (named joints + skeleton) per person
+- **Segmentation** — Instance segmentation: detect objects and their pixel masks
+- **Tracker** — Assign stable tracking ids to detections across frames
+- **Web Camera** — Starts a webcam stream and emits frame callbacks
+
+## Types
+- **Depth Frame** — Single-channel uint16 metric depth buffer (millimetres)
+- **Detection Result** — Object detections: bounding box, class, score (and tracking id)
+- **Gray Frame** — Single-channel uint8 luminance video frame (e.g. infrared)
+- **Landmark Result** — Landmark detections: a detection plus a set of landmark points
+- **Multiframe Callback** — Subscription for multi-frame streams
+- **Pose Result** — Human pose: landmarks with named joints and skeleton connections
+- **RGB Frame** — 3-channel uint8 colour video frame
+- **Segmentation Result** — Instance segmentation: detection plus a per-instance mask
+- **Vision Result** — A list of estimator results (base type for all result kinds)
+
+## Widgets
+- **NumpyViewerWidget** — Streaming video viewer for numpy arrays using custom StreamingViewer
+
+## Adapters
+- **GrayToRgbAdapter** — Replicate a single-channel grey frame to a 3-channel colour frame
+- **RgbToGrayAdapter** — Convert a 3-channel colour frame to a single-channel grey frame (luminance)

haybale_visiongraph-0.0.26/haybale_visiongraph/OVERVIEW.md

@@ -0,0 +1,33 @@
+# Visiongraph
+
+Visiongraph library
+
+## Nodes
+### Vision
+- **Annotate Results** — Draw estimator results (boxes / masks / poses) onto a frame
+- **Frame Event** — Triggered when a camera frame is ready; exposes colour/depth/infrared streams
+- **Frame Info Display** — Displays information about frames with live preview
+- **OAK-D Camera** — Opens an OAK-D depth camera and emits colour/depth/infrared frame callbacks
+- **Object Detector** — Detect objects in a frame (bounding box + class + score)
+- **Pose Estimator** — Estimate human body pose (named joints + skeleton) per person
+- **Segmentation** — Instance segmentation: detect objects and their pixel masks
+- **Tracker** — Assign stable tracking ids to detections across frames
+- **Web Camera** — Starts a webcam stream and emits frame callbacks
+
+## Types
+- **Depth Frame** — Single-channel uint16 metric depth buffer (millimetres)
+- **Detection Result** — Object detections: bounding box, class, score (and tracking id)
+- **Gray Frame** — Single-channel uint8 luminance video frame (e.g. infrared)
+- **Landmark Result** — Landmark detections: a detection plus a set of landmark points
+- **Multiframe Callback** — Subscription for multi-frame streams
+- **Pose Result** — Human pose: landmarks with named joints and skeleton connections
+- **RGB Frame** — 3-channel uint8 colour video frame
+- **Segmentation Result** — Instance segmentation: detection plus a per-instance mask
+- **Vision Result** — A list of estimator results (base type for all result kinds)
+
+## Widgets
+- **NumpyViewerWidget** — Streaming video viewer for numpy arrays using custom StreamingViewer
+
+## Adapters
+- **GrayToRgbAdapter** — Replicate a single-channel grey frame to a 3-channel colour frame
+- **RgbToGrayAdapter** — Convert a 3-channel colour frame to a single-channel grey frame (luminance)

haybale_visiongraph-0.0.26/haybale_visiongraph/QUICKREF.md

@@ -0,0 +1,30 @@
+# visiongraph — component index (v0.0.26)
+
+## node
+- `visiongraph:node:AnnotateNode` — Annotate Results — Draw estimator results (boxes / masks / poses) onto a frame  _tags: annotate, draw, overlay, render, result, visualize_
+- `visiongraph:node:FrameDisplayNode` — Frame Info Display — Displays information about frames with live preview  _tags: rgb, ir, frame, camera, display, preview, video, stream_
+- `visiongraph:node:NumpyFrameEventNode` — Frame Event — Triggered when a camera frame is ready; exposes colour/depth/infrared streams  _tags: 3d, depth, camera, oak, kinect, realsense, frame, event, rgb, ir, image_
+- `visiongraph:node:OakDCameraNode` — OAK-D Camera — Opens an OAK-D depth camera and emits colour/depth/infrared frame callbacks  _tags: oak, oak-d, depthai, luxonis, depth, camera, 3d, stream_
+- `visiongraph:node:ObjectDetectorNode` — Object Detector — Detect objects in a frame (bounding box + class + score)  _tags: object, detection, detector, yolo, deim, ssd, bbox, coco_
+- `visiongraph:node:PoseEstimatorNode` — Pose Estimator — Estimate human body pose (named joints + skeleton) per person  _tags: pose, human, body, skeleton, joints, landmark, mediapipe, movenet_
+- `visiongraph:node:SegmentationNode` — Segmentation — Instance segmentation: detect objects and their pixel masks  _tags: segmentation, instance, mask, yolo, maskrcnn, yolact_
+- `visiongraph:node:TrackerNode` — Tracker — Assign stable tracking ids to detections across frames  _tags: tracker, tracking, track, centroid, flate, motpy, id_
+- `visiongraph:node:WebCameraNode` — Web Camera — Starts a webcam stream and emits frame callbacks  _tags: webcam, camera, video, capture, stream_
+
+## type
+- `visiongraph:type:DEPTH_FRAME` — Depth Frame — Single-channel uint16 metric depth buffer (millimetres)
+- `visiongraph:type:DETECTION_RESULT` — Detection Result — Object detections: bounding box, class, score (and tracking id)
+- `visiongraph:type:GRAY_FRAME` — Gray Frame — Single-channel uint8 luminance video frame (e.g. infrared)
+- `visiongraph:type:LANDMARK_RESULT` — Landmark Result — Landmark detections: a detection plus a set of landmark points
+- `visiongraph:type:MULTIFRAME_CALLBACK` — Multiframe Callback — Subscription for multi-frame streams
+- `visiongraph:type:POSE_RESULT` — Pose Result — Human pose: landmarks with named joints and skeleton connections
+- `visiongraph:type:RGB_FRAME` — RGB Frame — 3-channel uint8 colour video frame
+- `visiongraph:type:SEGMENTATION_RESULT` — Segmentation Result — Instance segmentation: detection plus a per-instance mask
+- `visiongraph:type:VISION_RESULT` — Vision Result — A list of estimator results (base type for all result kinds)
+
+## adapter
+- `visiongraph:adapter:GrayToRgbAdapter` — GrayToRgbAdapter — Replicate a single-channel grey frame to a 3-channel colour frame
+- `visiongraph:adapter:RgbToGrayAdapter` — RgbToGrayAdapter — Convert a 3-channel colour frame to a single-channel grey frame (luminance)
+
+## widget
+- `visiongraph:widget:NumpyViewerWidget` — NumpyViewerWidget — Streaming video viewer for numpy arrays using custom StreamingViewer

haybale_visiongraph-0.0.26/haybale_visiongraph/__init__.py

@@ -0,0 +1,80 @@
+"""
+Visiongraph Library for Haywire
+"""
+
+from importlib.metadata import version as _pkg_version
+from pathlib import Path
+
+from haywire.core.library.base import BaseLibrary
+from haywire.core.library.compatibility import CompatibilityWarning
+from haywire.core.library.decorator import library
+from haywire.core.adapter.registry import AdapterRegistry
+from haywire.core.node.registry import NodeRegistry
+from haywire.core.types.registry import TypeRegistry
+
+from haywire.ui.skin.registry import SkinRegistry
+from haywire.ui.widget.registry import WidgetRegistry
+
+
+@library(
+    label="Visiongraph",
+    id="visiongraph",
+    version=_pkg_version("haybale-visiongraph"),
+    description="Visiongraph library",
+    url="https://github.com/haywire/haywire-repo/libraries/haybale-visiongraph",
+    help_url="https://docs.github.io/haywire_library",
+    author="Florian Briggisser, Martin Fröhlich",
+    author_url="https://author_url",
+    dependencies=['haybale_core'],
+    tags=['vision', 'camera', 'video', 'opencv'],
+    needs_refresh=True,
+    file_watcher=False,
+)
+class Library(BaseLibrary):
+    """Example library implementation"""
+
+    def register_components(self):
+        """Register all test components with the global registries"""
+
+        """Register nodes and types"""
+        base_path = Path(__file__).parent
+
+        # Register types (both variants and custom types)
+        self.add_folder_to_registry(folder_path=str(base_path / "types"), registry_cls=TypeRegistry)
+
+        # Register adapters
+        self.add_folder_to_registry(folder_path=str(base_path / "adapters"), registry_cls=AdapterRegistry)
+
+        # Register widgets
+        self.add_folder_to_registry(folder_path=str(base_path / "widgets"), registry_cls=WidgetRegistry)
+
+        # Register skins (node skins)
+        self.add_folder_to_registry(folder_path=str(base_path / "skins"), registry_cls=SkinRegistry)
+
+        # Register nodes
+        self.add_folder_to_registry(folder_path=str(base_path / "nodes"), registry_cls=NodeRegistry)
+
+    def validate(self) -> bool:
+        """Validate that the test library is properly structured"""
+        return True
+
+    def compatibility_warnings(self) -> list[CompatibilityWarning]:
+        """Append-only history of compatibility notices. See ADR 0005."""
+        from .nodes import FrameDisplayNode
+
+        return [
+            CompatibilityWarning(
+                version="0.0.13",
+                component=FrameDisplayNode,
+                message=(
+                    "The 'frame' inlet's widget visibility (show_widget) is now "
+                    "author-declared (WHEN_LINKED). Graphs saved before 0.0.13 may "
+                    "not show the live preview widget even when the inlet is linked. "
+                    "Reset the node to re-derive it from current code."
+                ),
+            ),
+        ]
+
+
+# Export for entry point discovery
+__all__ = ["Library"]

haybale_visiongraph-0.0.26/haybale_visiongraph/adapters/frame_adapters.py

@@ -0,0 +1,56 @@
+"""
+Frame adapters — conversions among the *image* frame types only.
+
+`GRAY_FRAME <-> RGB_FRAME` are registered because they are always safe to apply
+invisibly (channel replicate / luminance). **No adapter touches `DEPTH_FRAME`**:
+depth colourization is lossy, parameterized, and display-only, so it must be a
+deliberate node, never an auto edge-time coercion (see notes.md Q5).
+
+Adapters operate on **unwrapped values** — here, ``BaseFrame`` instances.
+"""
+
+import numpy as np
+from typing_extensions import override
+
+from haywire.core.adapter.base import BaseAdapter, adapter
+
+from ..types.frame_type import GRAY_FRAME, RGB_FRAME
+
+
+@adapter(
+    description="Replicate a single-channel grey frame to a 3-channel colour frame",
+    converts_from=GRAY_FRAME,
+    converts_to=RGB_FRAME,
+)
+class GrayToRgbAdapter(BaseAdapter):
+    """GRAY_FRAME -> RGB_FRAME by replicating the single channel across BGR."""
+
+    @override
+    def convert(self, value: GRAY_FRAME) -> RGB_FRAME:
+        rgb = value.data
+        if rgb is not None and rgb.ndim == 2:
+            rgb = np.stack([rgb, rgb, rgb], axis=-1)
+        return RGB_FRAME(data=rgb, timestamp=value.timestamp, frame_number=value.frame_number)
+
+    def get_test_value(self) -> GRAY_FRAME:
+        return GRAY_FRAME(data=np.zeros((4, 4), dtype=np.uint8))
+
+
+@adapter(
+    description="Convert a 3-channel colour frame to a single-channel grey frame (luminance)",
+    converts_from=RGB_FRAME,
+    converts_to=GRAY_FRAME,
+)
+class RgbToGrayAdapter(BaseAdapter):
+    """RGB_FRAME -> GRAY_FRAME via luminance."""
+
+    @override
+    def convert(self, value: RGB_FRAME) -> GRAY_FRAME:
+        gray = value.data
+        if gray is not None and gray.ndim == 3:
+            # Rec. 601 luma weights over BGR (OpenCV channel order).
+            gray = (gray[..., 0] * 0.114 + gray[..., 1] * 0.587 + gray[..., 2] * 0.299).astype(np.uint8)
+        return GRAY_FRAME(data=gray, timestamp=value.timestamp, frame_number=value.frame_number)
+
+    def get_test_value(self) -> RGB_FRAME:
+        return RGB_FRAME(data=np.zeros((4, 4, 3), dtype=np.uint8))

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.adapter.GrayToRgbAdapter.md

@@ -0,0 +1,15 @@
+# GrayToRgbAdapter
+
+`visiongraph:adapter:GrayToRgbAdapter` · kind: adapter
+
+Replicate a single-channel grey frame to a 3-channel colour frame
+
+## Details
+
+- **converts_from**: `visiongraph:type:GRAY_FRAME`
+- **converts_to**: `visiongraph:type:RGB_FRAME`
+- **priority**: `0`
+
+## Notes
+
+GRAY_FRAME -> RGB_FRAME by replicating the single channel across BGR.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.adapter.RgbToGrayAdapter.md

@@ -0,0 +1,15 @@
+# RgbToGrayAdapter
+
+`visiongraph:adapter:RgbToGrayAdapter` · kind: adapter
+
+Convert a 3-channel colour frame to a single-channel grey frame (luminance)
+
+## Details
+
+- **converts_from**: `visiongraph:type:RGB_FRAME`
+- **converts_to**: `visiongraph:type:GRAY_FRAME`
+- **priority**: `0`
+
+## Notes
+
+RGB_FRAME -> GRAY_FRAME via luminance.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.AnnotateNode.md

@@ -0,0 +1,40 @@
+# Annotate Results
+
+`visiongraph:node:AnnotateNode` · kind: node
+
+Draw estimator results (boxes / masks / poses) onto a frame
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| execute | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| result | inlet | core:type:PooledType | Connect one or more estimator result outlets |
+| frame | inlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| min_score | config | builtin:type:FLOAT | Decimal numberer |
+| show_info | config | builtin:type:BOOL | True or False |
+| frame_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| annotated | outlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+
+## Settings
+
+| name | bag | default | description |
+|---|---|---|---|
+| show_bounding_box | style | False | Draw the bounding box for landmark/pose results |
+| marker_size | style | 3 | Landmark marker radius in pixels |
+| stroke_width | style | 2 | Line width for boxes and skeleton connections |
+
+## Notes
+
+Overlay pooled vision results onto a frame.
+
+Inputs:
+    execute: Control flow in.
+    result: Pooled result lists (any mix of VISION_RESULT subtypes).
+    frame: The image to draw on (RGB_FRAME).
+    min_score: Hide results / landmarks below this confidence.
+    show_info: Draw labels / info text.
+
+Outputs:
+    frame_ready: Control flow out.
+    frame: The annotated frame (a copy; inputs are not mutated).

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.FrameDisplayNode.md

@@ -0,0 +1,36 @@
+# Frame Info Display
+
+`visiongraph:node:FrameDisplayNode` · kind: node
+
+Displays information about frames with live preview
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| execute | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| frame | inlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| info_display | config | builtin:type:STRING | Text data |
+| frame_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| frame_pass | outlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| timestamp | outlet | builtin:type:FLOAT | Decimal numberer |
+| frame_number | outlet | builtin:type:INT | Whole number |
+| width | outlet | builtin:type:INT | Whole number |
+| height | outlet | builtin:type:INT | Whole number |
+
+## Notes
+
+Displays frame information and live preview.
+
+Shows frame metadata and streams the video to an embedded viewer.
+
+Inputs:
+    execute: Control flow in
+    frame: Frame to display (RGB_FRAME type)
+
+Outputs:
+    frame_ready: Control flow out
+    timestamp: Time since stream start
+    frame_number: Sequential frame number
+    width: Frame width
+    height: Frame height

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.NumpyFrameEventNode.md

@@ -0,0 +1,31 @@
+# Frame Event
+
+`visiongraph:node:NumpyFrameEventNode` · kind: node
+
+Triggered when a camera frame is ready; exposes colour/depth/infrared streams
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| enable_rgb | config | builtin:type:BOOL | True or False |
+| enable_depth | config | builtin:type:BOOL | True or False |
+| enable_ir | config | builtin:type:BOOL | True or False |
+| subscription | outlet | visiongraph:type:MULTIFRAME_CALLBACK | Subscribe for camera frames |
+| frame_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| timestamp | outlet | builtin:type:FLOAT | Decimal numberer |
+| frame_number | outlet | builtin:type:INT | Whole number |
+| rgb | outlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+
+## Notes
+
+Event node that receives 3D-camera frame callbacks.
+
+Config:
+    rgb / depth / ir: Toggle which streams this node requests and exposes.
+
+Outputs:
+    subscription: MULTIFRAME_CALLBACK carrying the event name + requirements.
+    frame_ready: Control flow when a frame arrives.
+    rgb / depth / ir: The requested frame streams (dynamic outlets).
+    frame_number / timestamp: Frame metadata.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.OakDCameraNode.md

@@ -0,0 +1,80 @@
+# OAK-D Camera
+
+`visiongraph:node:OakDCameraNode` · kind: node
+
+Opens an OAK-D depth camera and emits colour/depth/infrared frame callbacks
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| start | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| stop | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| callbacks | inlet | core:type:PooledType | Connect to a Frame Event node |
+| status | config | builtin:type:STRING | Text data |
+| visiongraph.node.OakDCameraNode.device.mxid | config | builtin:type:STRING | Leave empty to auto-select the first available OAK device. |
+| started | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| stopped | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+
+## Settings
+
+| name | bag | default | description |
+|---|---|---|---|
+| mxid | device | '' | Leave empty to auto-select the first available OAK device. |
+| want_rgb | stream_flags | False | Set by connected subscribers. Not user-editable. |
+| want_depth | stream_flags | False | Set by connected subscribers. Not user-editable. |
+| want_ir | stream_flags | False | Set by connected subscribers. Not user-editable. |
+| enable_auto_exposure | color | True | Enables or disables auto exposure. Inert if Color is off. |
+| exposure | color | 20000 | Manual exposure time. Only applied while Auto Exposure is off. Inert if Color is off. |
+| iso | color | 800 | Sensor ISO sensitivity. Inert if Color is off. |
+| auto_exposure_compensation | color | 0 | Bias applied to auto exposure, in the range [-9, 9]. Inert if Color is off. |
+| enable_auto_white_balance | color | True | Enables or disables auto white balance. Inert if Color is off. |
+| white_balance | color | 4000 | Manual white balance in Kelvin. Only applied while Auto White Balance is off. Inert if Color is off. |
+| auto_white_balance_mode | color | 'AUTO' | Auto white balance scene preset (e.g. Daylight, Fluorescent). Inert if Color is off. |
+| auto_focus | color | True | Enables or disables auto focus. Inert if Color is off. |
+| focus_distance | color | 0 | Manual focus position. Only applied while Auto Focus is off. Inert if Color is off. |
+| brightness | color | 0 | Image brightness, in the range [-10, 10]. Inert if Color is off. |
+| contrast | color | 0 | Image contrast, in the range [-10, 10]. Inert if Color is off. |
+| saturation | color | 0 | Image saturation, in the range [-10, 10]. Inert if Color is off. |
+| sharpness | color | 0 | Image sharpness, in the range [0, 4]. Inert if Color is off. |
+| luma_denoise | color | 0 | Luminance noise reduction, in the range [0, 4]. Inert if Color is off. |
+| chroma_denoise | color | 0 | Chrominance (color) noise reduction, in the range [0, 4]. Inert if Color is off. |
+| anti_banding_mode | color | 'AUTO' | Reduces flicker banding from artificial lighting. Inert if Color is off. |
+| effect_mode | color | 'OFF' | Built-in color effect (e.g. Mono, Sepia, Negative). Inert if Color is off. |
+| laser_intensity | ir | 0.0 | Sets intensity of laser dot projector |
+| flood_intensity | ir | 0.0 | Sets intensity of the IR flood light |
+| preset_mode | depth | 'HIGH_DENSITY' | Stereo depth quality/speed preset. Requires a device restart to apply. |
+| median_filter | depth | 'KERNEL_7x7' | Disparity smoothing kernel. Requires a device restart to apply. |
+| left_right_check | depth | True | Better handling of occlusions. Requires a device restart to apply. |
+| subpixel | depth | False | Fractional disparity for longer-range accuracy. Restart required. |
+| extended_disparity | depth | False | Closer-in minimum depth, doubled disparity range. Restart required. |
+| frame_alignment | depth | 'Color' | Align the depth map to Color, Infrared, or leave Disabled. Restart required. |
+
+## Notes
+
+Starts an OAK-D capture stream in a separate thread and emits one callback
+per frame carrying every active stream.
+
+Inputs:
+    start: Open the device and begin capturing.
+    stop: Stop capturing and close the device.
+    callbacks: Pooled MULTIFRAME_CALLBACK subscriptions from event nodes.
+
+Outputs:
+    started: Triggered when the device opens successfully.
+    stopped: Triggered when capture stops.
+
+Settings:
+    device: Device MXID / name (empty = first available). Panel-only —
+        its dropdown resolves options via a live callable, which is only
+        safe on a setting() field (see `device.mxid` for why).
+    depth: Pipeline-construction params (read once at setup(), immutable
+        while running). Panel-only — not promotable to a port.
+    ir: Live IR laser/flood intensity. Promotable to an inlet — pushed
+        straight to the running device.
+    color: Live color-sensor controls (exposure/WB/focus/tuning).
+        Promotable to an inlet. Inert while enable_color is False (no
+        color control queue exists on the device in that case).
+    stream_flags: Read-only display of the requirement union gathered
+        from pooled subscribers (hb_gather_requirements). Not
+        user-editable, never promotable — set only via the callback edge.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.ObjectDetectorNode.md

@@ -0,0 +1,22 @@
+# Object Detector
+
+`visiongraph:node:ObjectDetectorNode` · kind: node
+
+Detect objects in a frame (bounding box + class + score)
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| execute | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| frame | inlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| model | config | builtin:type:STRING | Text data |
+| min_score | config | builtin:type:FLOAT | Decimal numberer |
+| status | config | builtin:type:STRING | Text data |
+| result_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| result | outlet | visiongraph:type:DETECTION_RESULT | Object detections: bounding box, class, score (and tracking id) |
+| count | outlet | builtin:type:INT | Whole number |
+
+## Notes
+
+Object detection family node — outlets ``DETECTION_RESULT``.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.PoseEstimatorNode.md

@@ -0,0 +1,22 @@
+# Pose Estimator
+
+`visiongraph:node:PoseEstimatorNode` · kind: node
+
+Estimate human body pose (named joints + skeleton) per person
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| execute | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| frame | inlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| model | config | builtin:type:STRING | Text data |
+| min_score | config | builtin:type:FLOAT | Decimal numberer |
+| status | config | builtin:type:STRING | Text data |
+| result_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| result | outlet | visiongraph:type:POSE_RESULT | Human pose: landmarks with named joints and skeleton connections |
+| count | outlet | builtin:type:INT | Whole number |
+
+## Notes
+
+Human-pose family node — outlets ``POSE_RESULT``.

haybale_visiongraph-0.0.26/haybale_visiongraph/docs/visiongraph.node.SegmentationNode.md

@@ -0,0 +1,22 @@
+# Segmentation
+
+`visiongraph:node:SegmentationNode` · kind: node
+
+Instance segmentation: detect objects and their pixel masks
+
+## Ports
+
+| id | direction | type | description |
+|---|---|---|---|
+| execute | inlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| frame | inlet | visiongraph:type:RGB_FRAME | 3-channel uint8 colour video frame |
+| model | config | builtin:type:STRING | Text data |
+| min_score | config | builtin:type:FLOAT | Decimal numberer |
+| status | config | builtin:type:STRING | Text data |
+| result_ready | outlet | core:type:EXEC | Signal for controlling execution flow between nodes |
+| result | outlet | visiongraph:type:SEGMENTATION_RESULT | Instance segmentation: detection plus a per-instance mask |
+| count | outlet | builtin:type:INT | Whole number |
+
+## Notes
+
+Instance-segmentation family node — outlets ``SEGMENTATION_RESULT``.