Class ProcessorContext
Defined in File holoinfer.hpp
-
class ProcessorContext
Processor Context class
Public Functions
-
ProcessorContext()
-
InferStatus initialize(const MultiMappings &process_operations, const Mappings &custom_kernels, bool use_cuda_graphs, const std::string config_path)
Initialize the preprocessor context
- Parameters
process_operations – Map of tensor name as key, mapped to list of operations to be applied in sequence on the tensor
custom_kernels – Map of custom kernel identifier, mapped to related value as a string
use_cuda_graphs – Flag to enable CUDA Graphs for processing custom CUDA kernels
config_path – Configuration path as a string
- Returns
InferStatus with appropriate holoinfer_code and message.
-
InferStatus process(const MultiMappings &tensor_oper_map, const MultiMappings &in_out_tensor_map, DataMap &processed_result_map, const std::map<std::string, std::vector<int>> &dimension_map, bool process_with_cuda, cudaStream_t cuda_stream = 0)
Process the tensors with operations as initialized. Toolkit supports one tensor input and output per model
- Parameters
tensor_oper_map – Map of tensor name as key, mapped to list of operations to be applied in sequence on the tensor
in_out_tensor_map – Map of input tensor name mapped to vector of output tensor names after processing
processed_result_map – Map is updated with output tensor name as key mapped to processed output as a vector of float32 type
dimension_map – Map is updated with model name as key mapped to dimension of processed data as a vector
process_with_cuda – Flag defining if processing should be done with CUDA
cuda_stream – CUDA stream to use when procseeing is done with CUDA
- Returns
InferStatus with appropriate holoinfer_code and message.
-
DataMap get_processed_data() const
Get output data per Tensor Toolkit supports one output per Tensor, in float32 type
- Returns
Map of tensor name as key mapped to the output float32 type data as a vector
-
DimType get_processed_data_dims() const
Get output dimension per model Toolkit supports one output per model
- Returns
Map of model as key mapped to the output dimension (of processed data) as a vector
-
ProcessorContext()