public gxf::Codelet

Class Documentation

class Postprocessor : public gxf::Codelet

Segmentation model postproceessing Codelet converting inference output to class index.

This Codelet performs segmentation model postprocessing in CUDA. It takes in the output of inference, either with the final softmax layer (multiclass) or sigmoid (2-class), and emits a Tensor<uint8_t> that contains the highest probability class index. The class index can then be consumed downstream for visualization or other purposes. The inference output currently supported are either HWC or NCHW.

Public Functions

gxf_result_t start() override

gxf_result_t tick() override

gxf_result_t stop() override

gxf_result_t registerInterface(gxf::Registrar *registrar) override

Class Postprocessor

Inheritance Relationships

Base Type

Class Documentation