DetectionSpec Struct

Detection-decoder metadata — tags tensors that carry packed detection output. More...

Declaration

struct simaai::neat::DetectionSpec { ... }

Included Headers

#include <TensorCore.h>

Public Member Attributes Index

std::string	format
	Wire-format token (e.g., "BBOX"). Consumer dispatches on this string. More...

Description

Detection-decoder metadata — tags tensors that carry packed detection output.

Set on tensors produced by detection-decoder stages (BoxDecode today). The format token identifies the wire layout the consumer should parse — for example "BBOX" for the standard uint32 count + N × 24-byte RawBox layout consumed by decode_bbox_tensor / pyneat.decode_bbox, or task-specific BoxDecode tokens for pose and segmentation payloads.

Historically this information was overloaded onto TessSpec::format, but tessellation is about MLA tile geometry, not about detection payload formats. DetectionSpec is the type-honest home for detection-stage output tags.

Definition at line 275 of file TensorCore.h.

Public Member Attributes

format

std::string simaai::neat::DetectionSpec::format

Wire-format token (e.g., "BBOX"). Consumer dispatches on this string.

Definition at line 277 of file TensorCore.h.

---

The documentation for this struct was generated from the following file:

• TensorCore.h

---

Generated via doxygen2docusaurus 2.0.0 by Doxygen 1.9.8.