SAMOptions

Options & Configuration · AiDotNet.VisionLanguage.Encoders

Configuration options for the Segment Anything Model (SAM) vision encoder.

Inheritance: ModelOptions → VisionEncoderOptions → SAMOptions

How It Works

SAM (Kirillov et al., 2023) consists of a ViT-based image encoder that produces image embeddings, which can be combined with prompt embeddings (points, boxes, masks) via a lightweight mask decoder for promptable segmentation of any object.

Constructors

Constructor Summary
SAMOptions() Initializes a new instance with default values.
SAMOptions(SAMOptions other) Initializes a new instance by copying from another instance.

Properties

Property Summary
bool UseRelativePositionalEncoding { get; set; } Gets or sets whether to use relative positional encoding (window attention).
int MaskDecoderDim { get; set; } Gets or sets the mask decoder embedding dimension.
int NumMaskDecoderLayers { get; set; } Gets or sets the number of mask decoder layers.
int NumMultimaskOutputs { get; set; } Gets or sets the number of multimask outputs.
int WindowSize { get; set; } Gets or sets the window size for windowed attention in the encoder.