NVIDIA DeepStream SDK API Reference

6.4 Release
NvDsInferParseDetectionParams Struct Reference

Detailed Description

Holds the detection parameters required for parsing objects.

Definition at line 191 of file nvdsinfer_custom_impl.h.

Data Fields

unsigned int numClassesConfigured
 Holds the number of classes requested to be parsed, starting with class ID 0. More...
 
std::vector< float > perClassPreclusterThreshold
 Holds a per-class vector of detection confidence thresholds to be applied prior to the clustering operation. More...
 
std::vector< float > perClassPostclusterThreshold
 
std::vector< float > & perClassThreshold = perClassPreclusterThreshold
 

Field Documentation

◆ numClassesConfigured

unsigned int NvDsInferParseDetectionParams::numClassesConfigured

Holds the number of classes requested to be parsed, starting with class ID 0.

Parsing functions may only output objects with class ID less than this value.

Definition at line 196 of file nvdsinfer_custom_impl.h.

◆ perClassPostclusterThreshold

std::vector<float> NvDsInferParseDetectionParams::perClassPostclusterThreshold

Definition at line 204 of file nvdsinfer_custom_impl.h.

◆ perClassPreclusterThreshold

std::vector<float> NvDsInferParseDetectionParams::perClassPreclusterThreshold

Holds a per-class vector of detection confidence thresholds to be applied prior to the clustering operation.

Parsing functions may only output an object with detection confidence greater than or equal to the vector element indexed by the object's class ID.

Definition at line 202 of file nvdsinfer_custom_impl.h.

◆ perClassThreshold

std::vector<float>& NvDsInferParseDetectionParams::perClassThreshold = perClassPreclusterThreshold
Deprecated:
Use perClassPreclusterThreshold instead.

Reference to maintain backward compatibility.

Definition at line 208 of file nvdsinfer_custom_impl.h.


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