ModelProfile Class Reference

Class for computing and managing various model statistics. More...

Public Member Functions
	ModelProfile (std::string model_id, std::string conf_path, int save_interval, int top_classes)
	Constructor to initialize ModelProfile object. More...
int	log_classification_model_stats (float inference_latency, const ClassificationResults &results)
	Logs statistics for a classification model. More...
int	log_embeddings (const std::vector< float > &embeddings)
	Logs model embeddings from the model. More...
int	log_embeddings (const std::vector< float > &embeddings, int cls)
	Logs Class wise model embeddings. More...
int	getNumDistributionBoxes () const
const distributionBox &	getDistributionBox (unsigned int index) const
void	registerStatistics ()

Public Attributes
frequent_class_sketch *	sketch1

Detailed Description

Class for computing and managing various model statistics.

Constructor & Destructor Documentation

ModelProfile()

ModelProfile::ModelProfile	(	std::string	model_id,
		std::string	conf_path,
		int	save_interval,
		int	top_classes
	)

Constructor to initialize ModelProfile object.

ModelProfile class constructor.

Parameters

model_id	Unique identifier for the model
no_of_classes	Number of classes the model predicts
saver	Reference to a Saver object used for saving model statistics

Member Function Documentation

log_classification_model_stats()

int ModelProfile::log_classification_model_stats	(	float	inference_latency,
		const ClassificationResults &	results
	)

Logs statistics for a classification model.

Logs classification model statistics.

Parameters

inference_latency	Time taken for model inference
results	Reference to the classification results

Returns

0 on success, negative value on error

Parameters

inference_latency	Time taken for model inference
results	Reference to the classification results

Returns

0 on success, negative value on error

This function iterates through the provided results and logs statistics for the most frequent classes. It updates the model_classes_stat map with scores for each class.

log_embeddings() [1/2]

int ModelProfile::log_embeddings

(

const std::vector< float > &

embeddings

)

Logs model embeddings from the model.

Logs embeddings.

Parameters

vector

of embeddings return 0 on sucess, negative value on error

log_embeddings() [2/2]

int ModelProfile::log_embeddings	(	const std::vector< float > &	embeddings,
		int	cls
	)

Logs Class wise model embeddings.

Computes and logs selected image statistics.

Parameters

vector	of embeddings
class	id return 0 on success , negative value on error
img	OpenCV image matrix
save_sample	Flag indicating whether to save samples exceeding thresholds

Returns

1 on success, error code on failure

---

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

• include/modelprofile.h
• src/profiles/modelprofile.cpp