NVIDIA DeepStream SDK API Reference

6.4 Release
GstNvInferTensorOutputObject Struct Reference

Detailed Description

Data type used for the refcounting and managing the usage of NvDsInferContext's batch output and the output buffers contained in it.

This is especially required when the tensor output is flowed along with buffers as metadata or when the segmentation output containing pointer to the NvDsInferContext allocated memory is attached to buffers as metadata. Whenever the last ref on the buffer is dropped, the callback to free the GstMiniObject-inherited GstNvInferTensorOutputObject is called and the batch_output can be released back to the NvDsInferContext.

Definition at line 132 of file gstnvinfer_impl.h.

Collaboration diagram for GstNvInferTensorOutputObject:

Data Fields

GstMiniObject mini_object
 Parent type. More...
 
NvDsInferContextPtr infer_context
 NvDsInferContext pointer which hold the resource. More...
 
NvDsInferContextBatchOutput batch_output
 NvDsInferContextBatchOutput instance whose output tensor buffers are being sent as meta data. More...
 

Field Documentation

◆ batch_output

NvDsInferContextBatchOutput GstNvInferTensorOutputObject::batch_output

NvDsInferContextBatchOutput instance whose output tensor buffers are being sent as meta data.

This batch output will be released back to the NvDsInferContext when the last ref on the mini_object is removed.

Definition at line 143 of file gstnvinfer_impl.h.

◆ infer_context

NvDsInferContextPtr GstNvInferTensorOutputObject::infer_context

NvDsInferContext pointer which hold the resource.

Definition at line 139 of file gstnvinfer_impl.h.

◆ mini_object

GstMiniObject GstNvInferTensorOutputObject::mini_object

Parent type.

Allows easy refcounting and destruction. Refcount will be increased by 1 for each frame/object for which NvDsInferTensorMeta will be generated.

Definition at line 137 of file gstnvinfer_impl.h.


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