Import Transfer Learning Toolkit Encoded ModelsΒΆ

group ee_nvdsinfer_tlt

Defines an API for importing Transfer Learning Toolkit encoded models.


bool NvDsInferCudaEngineGetFromTltModel(nvinfer1::IBuilder *const builder, const NvDsInferContextInitParams *const initParams, nvinfer1::DataType dataType, nvinfer1::ICudaEngine *&cudaEngine)

Decodes and creates a CUDA engine file from a TLT encoded model.

This function implements the NvDsInferCudaEngineGet interface. The correct key and model path must be provided in the tltModelKey and tltEncodedModelFilePath members of initParams. Other parameters applicable to UFF models also apply to TLT encoded models.