SAMModelSize

Enums · AiDotNet.Enums

Defines the backbone size variants for SAM (Segment Anything Model).

For Beginners

SAM is the original Segment Anything Model from Meta AI. It uses a Vision Transformer (ViT) backbone with varying sizes. Larger backbones produce more accurate masks but are slower. Choose ViTBase for a good balance, or ViTHuge for maximum accuracy.

How It Works

Technical Details: The ViT encoder processes images at 1024x1024 resolution using 16x16 patches. The model was trained on the SA-1B dataset containing 1B+ masks.

Reference: Kirillov et al., “Segment Anything”, ICCV 2023.

Fields

Field Summary
SAMModelSize ViTBase ViT-B backbone (91M params).
SAMModelSize ViTHuge ViT-H backbone (636M params).
SAMModelSize ViTLarge ViT-L backbone (308M params).
int value__