SSD

object SSD : ONNXModels.ObjectDetection<SSDObjectDetectionModel>

This model is a real-time neural network for object detection that detects 80 different classes (labels are available via org.jetbrains.kotlinx.dl.impl.dataset.Coco.V2014.labels method).

The model have an input with the shape is (1x3x1200x1200).

The model has 3 outputs:

boxes: (1x'nbox'x4)
labels: (1x'nbox')
scores: (1x'nbox')

See also

Detailed description of SSD model and its pre- and postprocessing in onnx/models repository.

Functions

model

jvm

open fun model(modelHub: ModelHub): OnnxInferenceModel

pretrainedModel

jvm

open override fun pretrainedModel(modelHub: ModelHub): SSDObjectDetectionModel

Properties

inputShape

jvm

open val inputShape: LongArray?

Shape of the input accepted by this model, without batch size.

modelRelativePath

jvm

open override val modelRelativePath: String

preprocessor

jvm

open override val preprocessor: Operation<Pair<FloatArray, TensorShape>, Pair<FloatArray, TensorShape>>