MetoakDepthOptions Struct
Metoak S315 SIMOR depth-map Node: I420->RGB + disparity->metric-depth/point-cloud, offloaded to the native simor_depth_map EV74 kernel (graph id 20). More...
Declaration
Included Headers
Public Member Functions Index
| std::string | summary () const |
|
Human-readable, non-throwing summary for logs, diagnostics, and Python repr. More... | |
Public Member Attributes Index
| int | width = 0 |
|
Even input width in [8, 2048]; use native sensor geometry (640 for S315). More... | |
| int | height = 0 |
|
Even input height in [8, 1536]; use native sensor geometry (360 for S315). More... | |
| int | debug = 0 |
|
EV graph debug level. Current native visual graphs accept values in [0,2]. More... | |
| int | num_buffers = 0 |
|
Optional processcvu buffer override. More... | |
| std::string | element_name |
|
Optional GStreamer/processcvu element name. Empty means Neat generates a stable name. More... | |
| std::string | y_name = "y_src" |
|
I420 Y input tensor name. More... | |
| std::string | u_name = "u_src" |
|
I420 U input tensor name. More... | |
| std::string | v_name = "v_src" |
|
I420 V input tensor name. More... | |
| std::string | disp_name = "disp_src" |
|
Raw disparity input tensor name. More... | |
| std::string | bf_mm_name = "bf_mm_src" |
|
Per-frame calibrated BF (base * focal length, mm) input tensor name. More... | |
| std::string | proj_name = "proj_src" |
|
Per-frame projection {fx_fy, cx, cy} input tensor name. More... | |
| std::string | rgb_output_name = "rgb_dst" |
|
RGB output tensor name. More... | |
| std::string | depth_output_name = "depth_dst" |
|
Metric depth output tensor name. More... | |
| std::string | points_output_name = "points_dst" |
|
XYZ point-cloud output tensor name. More... | |
Description
Metoak S315 SIMOR depth-map Node: I420->RGB + disparity->metric-depth/point-cloud, offloaded to the native simor_depth_map EV74 kernel (graph id 20).
Fixed six-input/three-output contract (batch 1; see the "Metoak EV74 support in Neat" migration notes for the full rationale). Unlike the other native visual graphs above, this one is not a feature/tracking kernel: bf_mm_name/proj_name are per-frame scalar calibration inputs (not batched images), and it has three heterogeneous outputs rather than one primary tensor.
Tensor names currently must retain their canonical defaults; unsupported aliases fail during contract compilation, before any device access.
Public ABI (fixed geometry from width/height; disparity subpixel scale 32 is baked into the kernel, not a parameter here):
- inputs:
- y_name, UInt8 [height,width] – I420 Y
- u_name, UInt8 [height/2,width/2] – I420 U
- v_name, UInt8 [height/2,width/2] – I420 V
- disp_name, UInt16 [height,width] – raw disparity
- bf_mm_name, Float32 [1] – selected calibration BF (mm)
- proj_name, Float32 [3] – {fx_fy, cx, cy}
- outputs:
- rgb_output_name, UInt8 [height,width,3] – interleaved RGB
- depth_output_name, UInt16 [height,width] – millimeters, 0 = invalid (primary)
- points_output_name, Float32 [height,width,3] – interleaved XYZ meters, NaN = invalid
depth_output_name is the primary output for output_spec()'s single-boundary description only; all three outputs are always published together – selecting one as primary must not be read as hiding the other two.
Definition at line 224 of file VisualFrontend.h.
Public Member Functions
summary()
|
Human-readable, non-throwing summary for logs, diagnostics, and Python repr.
Definition at line 257 of file VisualFrontend.h.
Public Member Attributes
bf_mm_name
|
Per-frame calibrated BF (base * focal length, mm) input tensor name.
Definition at line 246 of file VisualFrontend.h.
debug
|
EV graph debug level. Current native visual graphs accept values in [0,2].
Definition at line 230 of file VisualFrontend.h.
depth_output_name
|
Metric depth output tensor name.
Definition at line 252 of file VisualFrontend.h.
disp_name
|
Raw disparity input tensor name.
Definition at line 244 of file VisualFrontend.h.
element_name
|
Optional GStreamer/processcvu element name. Empty means Neat generates a stable name.
Definition at line 235 of file VisualFrontend.h.
height
|
Even input height in [8, 1536]; use native sensor geometry (360 for S315).
Definition at line 228 of file VisualFrontend.h.
num_buffers
|
Optional processcvu buffer override.
0 resolves to the async four-buffer default. Sync builds apply their normal pool clamp; async builds require four buffers.
Definition at line 233 of file VisualFrontend.h.
points_output_name
|
XYZ point-cloud output tensor name.
Definition at line 254 of file VisualFrontend.h.
proj_name
|
Per-frame projection {fx_fy, cx, cy} input tensor name.
Definition at line 248 of file VisualFrontend.h.
rgb_output_name
|
RGB output tensor name.
Definition at line 250 of file VisualFrontend.h.
u_name
|
I420 U input tensor name.
Definition at line 240 of file VisualFrontend.h.
v_name
|
I420 V input tensor name.
Definition at line 242 of file VisualFrontend.h.
width
|
Even input width in [8, 2048]; use native sensor geometry (640 for S315).
Definition at line 226 of file VisualFrontend.h.
y_name
|
I420 Y input tensor name.
Definition at line 238 of file VisualFrontend.h.
The documentation for this struct was generated from the following file:
Generated via doxygen2docusaurus 2.0.0 by Doxygen 1.9.8.