Mask2FormerOptions
Options & Configuration · AiDotNet.ComputerVision.Segmentation.Foundation
Configuration options for the Mask2Former universal segmentation model.
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → PanopticSegmentationOptions → Mask2FormerOptions
For Beginners
Mask2Former options inherit from NeuralNetworkOptions, which provides a Seed property for reproducibility. Mask2Former is a universal model that can perform semantic, instance, and panoptic segmentation with a single architecture by using masked cross-attention in its transformer decoder.
Constructors
| Constructor | Summary |
|---|---|
Mask2FormerOptions() |
Initializes a new instance with default values. |
Mask2FormerOptions(Mask2FormerOptions other) |
Initializes a new instance by copying from another instance. |
Properties
| Property | Summary |
|---|---|
Mask2FormerModelSize ModelSize { get; set; } |
Gets or sets the backbone size variant. |
double LearningRate { get; set; } |
Gets or sets the optimizer learning rate. |
double MaxGradientNorm { get; set; } |
Gets or sets the gradient-norm clipping threshold; 0 disables clipping. |
double WeightDecay { get; set; } |
Gets or sets the AdamW weight decay. |
Methods
| Method | Summary |
|---|---|
void Validate() |
Throws if a value this model requires has been left unset or is not positive. |