Detailed Description

Holds info about one frame in a batch for inferencing.

Definition at line 50 of file gstnvinfer_impl.h.

Collaboration diagram for GstNvInferFrame:

[legend]

Data Fields
gdouble	scale_ratio_x = 0
	Ratio by which the frame / object crop was scaled in the horizontal direction. More...

gdouble	scale_ratio_y = 0
	Ratio by which the frame / object crop was scaled in the vertical direction. More...

guint	offset_left = 0
	Offsets if symmetric padding was performed whille scaling objects to network resolition. More...

guint	offset_top = 0

guint	roi_left = 0
	roi left and top for preprocessed tensor More...

guint	roi_top = 0

NvDsObjectMeta *	obj_meta = nullptr
	NvDsObjectParams belonging to the object to be classified. More...

NvDsFrameMeta *	frame_meta = nullptr

NvDsRoiMeta *	roi_meta = nullptr

guint	batch_index = 0
	Index of the frame in the batched input GstBuffer. More...

gulong	frame_num = 0
	Frame number of the frame from the source. More...

NvBufSurfaceParams *	input_surf_params = nullptr
	The buffer structure the object / frame was converted from. More...

gpointer	converted_frame_ptr = nullptr
	Pointer to the converted frame memory. More...

std::weak_ptr< GstNvInferObjectHistory >	history
	Pointer to the structure holding inference history for the object. More...

Field Documentation

◆ batch_index

guint GstNvInferFrame::batch_index = 0

Index of the frame in the batched input GstBuffer.

Not required for classifiers.

Definition at line 72 of file gstnvinfer_impl.h.

◆ converted_frame_ptr

gpointer GstNvInferFrame::converted_frame_ptr = nullptr

Pointer to the converted frame memory.

This memory contains the frame converted to RGB/RGBA and scaled to network resolution. This memory is given to NvDsInferContext as input for pre-processing and inferencing.

Definition at line 80 of file gstnvinfer_impl.h.

◆ frame_meta

NvDsFrameMeta* GstNvInferFrame::frame_meta = nullptr

Definition at line 68 of file gstnvinfer_impl.h.

◆ frame_num

gulong GstNvInferFrame::frame_num = 0

Frame number of the frame from the source.

Definition at line 74 of file gstnvinfer_impl.h.

◆ history

std::weak_ptr<GstNvInferObjectHistory> GstNvInferFrame::history

Pointer to the structure holding inference history for the object.

Should be NULL when inferencing on frames.

Definition at line 83 of file gstnvinfer_impl.h.

◆ input_surf_params

NvBufSurfaceParams* GstNvInferFrame::input_surf_params = nullptr

The buffer structure the object / frame was converted from.

Definition at line 76 of file gstnvinfer_impl.h.

◆ obj_meta

NvDsObjectMeta* GstNvInferFrame::obj_meta = nullptr

NvDsObjectParams belonging to the object to be classified.

Definition at line 67 of file gstnvinfer_impl.h.

◆ offset_left

guint GstNvInferFrame::offset_left = 0

Offsets if symmetric padding was performed whille scaling objects to network resolition.

Definition at line 61 of file gstnvinfer_impl.h.

◆ offset_top

guint GstNvInferFrame::offset_top = 0

Definition at line 62 of file gstnvinfer_impl.h.

◆ roi_left

guint GstNvInferFrame::roi_left = 0

roi left and top for preprocessed tensor

Definition at line 64 of file gstnvinfer_impl.h.

◆ roi_meta

NvDsRoiMeta* GstNvInferFrame::roi_meta = nullptr

Definition at line 69 of file gstnvinfer_impl.h.

◆ roi_top

guint GstNvInferFrame::roi_top = 0

Definition at line 65 of file gstnvinfer_impl.h.

◆ scale_ratio_x

gdouble GstNvInferFrame::scale_ratio_x = 0

Ratio by which the frame / object crop was scaled in the horizontal direction.

Required when scaling the detector boxes from the network resolution to input resolution. Not required for classifiers.

Definition at line 54 of file gstnvinfer_impl.h.

◆ scale_ratio_y

gdouble GstNvInferFrame::scale_ratio_y = 0

Ratio by which the frame / object crop was scaled in the vertical direction.