SparseBox3DDecoder Class Reference

Detailed Description

Definition at line 39 of file sources/gst-plugins/gst-nvdsvideotemplate/ds-sparse4d/postprocess/decoder.h.

Public Member Functions
	SparseBox3DDecoder (int num_output=300, float score_threshold=0.0f, bool sorted=true, torch::Tensor group_origin=torch::Tensor(), int gpu_id=0, PrettyLogger *logger=nullptr)
	Constructor for SparseBox3DDecoder. More...
torch::Tensor	decode_box (torch::Tensor &box, bool dummy_rotation=true)
	Decodes box parameters from network output format to actual box dimensions. More...
std::vector< DecodedResult >	decode (const std::vector< torch::Tensor > &cls_scores, const std::vector< torch::Tensor > &box_preds, const torch::Tensor &instance_id=torch::Tensor(), const torch::Tensor &instance_feature=torch::Tensor(), const std::vector< torch::Tensor > &quality=std::vector< torch::Tensor >(), int output_idx=-1)
	Decodes model predictions into final detection outputs. More...
	SparseBox3DDecoder (int num_output=300, float score_threshold=0.0f, bool sorted=true, torch::Tensor group_origin=torch::Tensor(), int gpu_id=0, PrettyLogger *logger=nullptr)
	Constructor for SparseBox3DDecoder. More...
torch::Tensor	decode_box (torch::Tensor &box, bool dummy_rotation=true)
	Decodes box parameters from network output format to actual box dimensions. More...
std::vector< DecodedResult >	decode (const std::vector< torch::Tensor > &cls_scores, const std::vector< torch::Tensor > &box_preds, const torch::Tensor &instance_id=torch::Tensor(), const torch::Tensor &instance_feature=torch::Tensor(), const std::vector< torch::Tensor > &quality=std::vector< torch::Tensor >(), int output_idx=-1)
	Decodes model predictions into final detection outputs. More...

Constructor & Destructor Documentation

SparseBox3DDecoder() [1/2]

SparseBox3DDecoder::SparseBox3DDecoder	(	int	num_output = 300,
		float	score_threshold = 0.0f,
		bool	sorted = true,
		torch::Tensor	group_origin = torch::Tensor(),
		int	gpu_id = 0,
		PrettyLogger *	logger = nullptr
	)

Constructor for SparseBox3DDecoder.

Parameters

num_output	Maximum number of detections to output
score_threshold	Confidence threshold for filtering detections
sorted	Whether to sort the results by score
group_origin	Group origin values
gpu_id	GPU device ID to use
logger	Logger instance for debug messages

SparseBox3DDecoder() [2/2]

SparseBox3DDecoder::SparseBox3DDecoder	(	int	num_output = 300,
		float	score_threshold = 0.0f,
		bool	sorted = true,
		torch::Tensor	group_origin = torch::Tensor(),
		int	gpu_id = 0,
		PrettyLogger *	logger = nullptr
	)

Constructor for SparseBox3DDecoder.

Parameters

num_output	Maximum number of detections to output
score_threshold	Confidence threshold for filtering detections
sorted	Whether to sort the results by score
group_origin	Group origin values
gpu_id	GPU device ID to use
logger	Logger instance for debug messages

Member Function Documentation

decode() [1/2]

std::vector<DecodedResult> SparseBox3DDecoder::decode	(	const std::vector< torch::Tensor > &	cls_scores,
		const std::vector< torch::Tensor > &	box_preds,
		const torch::Tensor &	instance_id = torch::Tensor(),
		const torch::Tensor &	instance_feature = torch::Tensor(),
		const std::vector< torch::Tensor > &	quality = std::vector< torch::Tensor >(),
		int	output_idx = -1
	)

Decodes model predictions into final detection outputs.

Parameters

cls_scores	Classification scores from the model
box_preds	Box predictions from the model
instance_id	Instance IDs (optional)
instance_feature	Instance features (optional)
quality	Quality scores including centerness (optional)
output_idx	Output index to process

Returns

Vector of DecodedResult structures with detection results

decode() [2/2]

std::vector<DecodedResult> SparseBox3DDecoder::decode	(	const std::vector< torch::Tensor > &	cls_scores,
		const std::vector< torch::Tensor > &	box_preds,
		const torch::Tensor &	instance_id = torch::Tensor(),
		const torch::Tensor &	instance_feature = torch::Tensor(),
		const std::vector< torch::Tensor > &	quality = std::vector< torch::Tensor >(),
		int	output_idx = -1
	)

Decodes model predictions into final detection outputs.

Parameters

cls_scores	Classification scores from the model
box_preds	Box predictions from the model
instance_id	Instance IDs (optional)
instance_feature	Instance features (optional)
quality	Quality scores including centerness (optional)
output_idx	Output index to process

Returns

Vector of DecodedResult structures with detection results

decode_box() [1/2]

torch::Tensor SparseBox3DDecoder::decode_box	(	torch::Tensor &	box,
		bool	dummy_rotation = true
	)

Decodes box parameters from network output format to actual box dimensions.

Parameters

box	Input tensor containing encoded box parameters
dummy_rotation	If true, adds dummy x and y rotation values

Returns

Decoded box parameters

decode_box() [2/2]

torch::Tensor SparseBox3DDecoder::decode_box	(	torch::Tensor &	box,
		bool	dummy_rotation = true
	)

Decodes box parameters from network output format to actual box dimensions.

Parameters

box	Input tensor containing encoded box parameters
dummy_rotation	If true, adds dummy x and y rotation values

Returns

Decoded box parameters

---

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

• sources/gst-plugins/gst-nvdsvideotemplate/ds-sparse4d/postprocess/decoder.h