NvDsSampleModelsExt

Components for using the sample models provided as part of DeepStreamSDK.

  • UUID: 909dc1ea-c654-44c1-97a3-0b8fec12141a

  • Version: 0.0.4

  • Author: NVIDIA

  • License: Proprietary

Components

nvidia::deepstream::NvDsResnet10_4ClassDetectorModel

4-class primary model detecting vehicles, persons, bicycles and roadsigns. Contains the caffemodel and prototxt, labels file, INT8 calibration file and the nvinfer configuration file for the model.

  • Component ID: c2bed4f0-b9dd-43ab-b693-e0cbdd4f96e8

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

cluster-mode

Object clustering mode to use.

Valid values:

1: DBSCAN

2: NMS

3: DBSCAN + NMS (Hybrid)

4: No clustering

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 2


dbscan-min-score

Minimum sum of confidence of all the neighbors in a cluster for it to be considered a valid cluster.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.0


detected-max-h

Maximum height in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


detected-max-w

Maximum width in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


detected-min-h

Minimum height in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


detected-min-w

Minimum width in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


eps

Epsilon values for DBSCAN clustering algorithm

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.7


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


input-object-max-height

When used in secondary mode only on objects with this maximum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-max-width

When used in secondary mode only on objects with this maximum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-min-height

When used in secondary mode only on objects with this minimum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


input-object-min-width

When used in secondary mode only on objects with this minimum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


min-boxes

Minimum number of points required to form a dense region for DBSCAN algorithm

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 3


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


nms-iou-threshold

Maximum IOU score between two proposals after which the proposal with the lower confidence will be rejected.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.5


post-cluster-threshold

Detection threshold to be applied post clustering operation for all classes

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.0


pre-cluster-threshold

Detection threshold to be applied prior to clustering operation for all classes

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.4


roi-bottom-offset

Offset of the RoI from the bottom of the frame. Only objects within the RoI are output.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


roi-top-offset

Offset of the RoI from the top of the frame. Only objects within the RoI are output.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-compute-hw

Compute hardware to use for scaling frames / object crops to network resolution.

Valid values:

0: Platform default - GPU (dGPU), VIC (Jetson)

1: GPU

2: VIC (Jetson only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-filter

The filter to use for scaling frames / object crops to network resolution.

Valid values:

0: Nearest

1: Bilinear

2: Cubic(GPU)/5-Tap(VIC)

3: Super(GPU)/10-Tap(VIC)

4: Lanzos(GPU)/Smart(VIC)

5: Nicest(VIC only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


topK

Keep only top K objects with highest detection scores.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 20


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0

nvidia::deepstream::NvDsSecondaryCarColorClassifierModel

Secondary car color classification model. Contains the caffemodel and prototxt, labels file, INT8 calibration file, the mean ppm file and the nvinfer configuration file for the model.

  • Component ID: f7a64e00-f01f-4eaf-8c94-d686c28b9e48

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

classifier-async-mode

Enables inference on detected objects and asynchronous metadata attachments.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: True


classifier-threshold

Minimum threshold label probability. The classifier outputs the label having the highest probability if it is greater than this threshold

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.51


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


input-object-max-height

When used in secondary mode only on objects with this maximum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-max-width

When used in secondary mode only on objects with this maximum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-min-height

When used in secondary mode only on objects with this minimum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


input-object-min-width

When used in secondary mode only on objects with this minimum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


scaling-compute-hw

Compute hardware to use for scaling frames / object crops to network resolution.

Valid values:

0: Platform default - GPU (dGPU), VIC (Jetson)

1: GPU

2: VIC (Jetson only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-filter

The filter to use for scaling frames / object crops to network resolution.

Valid values:

0: Nearest

1: Bilinear

2: Cubic(GPU)/5-Tap(VIC)

3: Super(GPU)/10-Tap(VIC)

4: Lanzos(GPU)/Smart(VIC)

5: Nicest(VIC only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0

nvidia::deepstream::NvDsSecondaryCarMakeClassifierModel

Secondary car make classification model. Contains the caffemodel and prototxt, labels file, INT8 calibration file, the mean ppm file and the nvinfer configuration file for the model.

  • Component ID: f10b9a8c-d8d4-4ed8-ad54-8d8dca2392b9

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

classifier-async-mode

Enables inference on detected objects and asynchronous metadata attachments.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: True


classifier-threshold

Minimum threshold label probability. The classifier outputs the label having the highest probability if it is greater than this threshold

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.51


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


input-object-max-height

When used in secondary mode only on objects with this maximum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-max-width

When used in secondary mode only on objects with this maximum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-min-height

When used in secondary mode only on objects with this minimum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


input-object-min-width

When used in secondary mode only on objects with this minimum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


scaling-compute-hw

Compute hardware to use for scaling frames / object crops to network resolution.

Valid values:

0: Platform default - GPU (dGPU), VIC (Jetson)

1: GPU

2: VIC (Jetson only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-filter

The filter to use for scaling frames / object crops to network resolution.

Valid values:

0: Nearest

1: Bilinear

2: Cubic(GPU)/5-Tap(VIC)

3: Super(GPU)/10-Tap(VIC)

4: Lanzos(GPU)/Smart(VIC)

5: Nicest(VIC only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0

nvidia::deepstream::NvDsSecondaryVehicleTypeClassifierModel

Secondary vehicle type classification model. Contains the caffemodel and prototxt, labels file, INT8 calibration file, the mean ppm file and the nvinfer configuration file for the model.

  • Component ID: 47728594-4a13-4460-a906-32c3ae52a61a

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

classifier-async-mode

Enables inference on detected objects and asynchronous metadata attachments.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: True


classifier-threshold

Minimum threshold label probability. The classifier outputs the label having the highest probability if it is greater than this threshold

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.51


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


input-object-max-height

When used in secondary mode only on objects with this maximum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-max-width

When used in secondary mode only on objects with this maximum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-min-height

When used in secondary mode only on objects with this minimum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


input-object-min-width

When used in secondary mode only on objects with this minimum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 128


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


scaling-compute-hw

Compute hardware to use for scaling frames / object crops to network resolution.

Valid values:

0: Platform default - GPU (dGPU), VIC (Jetson)

1: GPU

2: VIC (Jetson only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-filter

The filter to use for scaling frames / object crops to network resolution.

Valid values:

0: Nearest

1: Bilinear

2: Cubic(GPU)/5-Tap(VIC)

3: Super(GPU)/10-Tap(VIC)

4: Lanzos(GPU)/Smart(VIC)

5: Nicest(VIC only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0

nvidia::deepstream::NvDsSonyCAudioClassifierModel

Audio classifier trained on the SONYC urban noise database. Contains the onnx file, labels file and the nvinfer configuration file for the model.

  • Component ID: d8d8f07e-3502-4124-ac4f-15f521cf2fb9

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

classifier-threshold

Minimum threshold label probability. The classifier outputs the label having the highest probability if it is greater than this threshold

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.0


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0

nvidia::deepstream::NvDsCarDetector360dModel

Car detector model for 360d use case. Contains the caffemodel and prototxt, labels file, INT8 calibration file and the nvinfer configuration file for the model.

  • Component ID: 8e3defe9-faf4-4417-945f-4ceb3950ca20

  • Base Type: nvidia::deepstream::INvDsInferModelConfigComponent

Parameters

cluster-mode

Object clustering mode to use.

Valid values:

1: DBSCAN

2: NMS

3: DBSCAN + NMS (Hybrid)

4: No clustering

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


dbscan-min-score

Minimum sum of confidence of all the neighbors in a cluster for it to be considered a valid cluster.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.0


detected-max-h

Maximum height in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1920


detected-max-w

Maximum width in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1920


detected-min-h

Minimum height in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


detected-min-w

Minimum width in pixels of detected objects to be output by the detector

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


enable-dla

Enable DLA for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_BOOL

  • Default: False


eps

Epsilon values for DBSCAN clustering algorithm

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.9


infer-dims

Binding dimensions to set on the image input layer. Format - <dim1>;<dim2>;<dim3>…

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_STRING


input-object-max-height

When used in secondary mode only on objects with this maximum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-max-width

When used in secondary mode only on objects with this maximum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64


input-object-min-height

When used in secondary mode only on objects with this minimum height

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


input-object-min-width

When used in secondary mode only on objects with this minimum width

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


min-boxes

Minimum number of points required to form a dense region for DBSCAN algorithm

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 3


model-engine-file

Path to the model engine file. Absolute or relative to the extension directory.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FILE


network-mode

Data format to be used by inference.

Valid values:

0: FP32

1: INT8

2: FP16

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 1


nms-iou-threshold

Maximum IOU score between two proposals after which the proposal with the lower confidence will be rejected.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.5


post-cluster-threshold

Detection threshold to be applied post clustering operation for all classes

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.0


pre-cluster-threshold

Detection threshold to be applied prior to clustering operation for all classes

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_FLOAT64

  • Default: 0.5


roi-bottom-offset

Offset of the RoI from the bottom of the frame. Only objects within the RoI are output.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


roi-top-offset

Offset of the RoI from the top of the frame. Only objects within the RoI are output.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-compute-hw

Compute hardware to use for scaling frames / object crops to network resolution.

Valid values:

0: Platform default - GPU (dGPU), VIC (Jetson)

1: GPU

2: VIC (Jetson only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


scaling-filter

The filter to use for scaling frames / object crops to network resolution.

Valid values:

0: Nearest

1: Bilinear

2: Cubic(GPU)/5-Tap(VIC)

3: Super(GPU)/10-Tap(VIC)

4: Lanzos(GPU)/Smart(VIC)

5: Nicest(VIC only)

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0


tensor-meta-pool-size

Size of the output tensor meta pool

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 6


topK

Keep only top K objects with highest detection scores.

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 20


use-dla-core

DLA Core to use for inferencing

  • Flags: GXF_PARAMETER_FLAGS_OPTIONAL

  • Type: GXF_PARAMETER_TYPE_UINT64

  • Default: 0