EvalAccumulator Class Reference

Detailed Description

EvalAccumulator: per-epoch accumulator across all (role, model) pairs.

add_frame() is called from the GStreamer streaming thread. The engine serialises calls to this (one chain function per pad), so no locking needed.

Definition at line 77 of file sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h.

Public Member Functions
void	add_frame (const std::string &role, const std::string &model_name, const std::string &model_version, const std::vector< std::string > &preds_label, const std::vector< float > &preds_conf, const std::vector< float > &preds_x1, const std::vector< float > &preds_y1, const std::vector< float > &preds_x2, const std::vector< float > &preds_y2, const std::vector< GTBox > &gt_boxes)
	add_frame: record one frame's predictions vs GT. More...
void	reset ()
const std::map< ModelKey, ModelAccum > &	models () const
uint32_t	total_frames () const
void	add_frame (const std::string &role, const std::string &model_name, const std::string &model_version, const std::vector< std::string > &preds_label, const std::vector< float > &preds_conf, const std::vector< float > &preds_x1, const std::vector< float > &preds_y1, const std::vector< float > &preds_x2, const std::vector< float > &preds_y2, const std::vector< GTBox > &gt_boxes)
	add_frame: record one frame's predictions vs GT. More...
void	reset ()
const std::map< ModelKey, ModelAccum > &	models () const
uint32_t	total_frames () const

Member Function Documentation

add_frame() [1/2]

void EvalAccumulator::add_frame	(	const std::string &	role,
		const std::string &	model_name,
		const std::string &	model_version,
		const std::vector< std::string > &	preds_label,
		const std::vector< float > &	preds_conf,
		const std::vector< float > &	preds_x1,
		const std::vector< float > &	preds_y1,
		const std::vector< float > &	preds_x2,
		const std::vector< float > &	preds_y2,
		const std::vector< GTBox > &	gt_boxes
	)

add_frame: record one frame's predictions vs GT.

role / model_name: from InferenceProvenanceMeta (Mode B) or hardcoded (Mode A). preds_*: parallel prediction arrays from NvDsObjectMeta. gt_boxes: GT for (camera_id, frame_num) — empty when GT unavailable.

add_frame() [2/2]

void EvalAccumulator::add_frame	(	const std::string &	role,
		const std::string &	model_name,
		const std::string &	model_version,
		const std::vector< std::string > &	preds_label,
		const std::vector< float > &	preds_conf,
		const std::vector< float > &	preds_x1,
		const std::vector< float > &	preds_y1,
		const std::vector< float > &	preds_x2,
		const std::vector< float > &	preds_y2,
		const std::vector< GTBox > &	gt_boxes
	)

add_frame: record one frame's predictions vs GT.

role / model_name: from InferenceProvenanceMeta (Mode B) or hardcoded (Mode A). preds_*: parallel prediction arrays from NvDsObjectMeta. gt_boxes: GT for (camera_id, frame_num) — empty when GT unavailable.

models() [1/2]

const std::map<ModelKey, ModelAccum>& EvalAccumulator::models ( ) const

inline

Definition at line 100 of file 9.2/sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h.

models() [2/2]

const std::map<ModelKey, ModelAccum>& EvalAccumulator::models ( ) const

inline

Definition at line 100 of file sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h.

reset() [1/2]

void EvalAccumulator::reset

(

)

reset() [2/2]

void EvalAccumulator::reset

(

)

total_frames() [1/2]

uint32_t EvalAccumulator::total_frames ( ) const

inline

Definition at line 101 of file sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h.

total_frames() [2/2]

uint32_t EvalAccumulator::total_frames ( ) const

inline

Definition at line 101 of file 9.2/sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h.

---

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

• sources/gst-plugins/gst-nvinfereval/nvds_eval_accumulator.h