VisualFeatureType
Enums · AiDotNet.VisionLanguage.Encoders
Specifies the visual feature extraction method for foundational VLMs.
Fields
| Field | Summary |
|---|---|
VisualFeatureType GridFeatures |
Grid features from a CNN backbone. |
VisualFeatureType PatchEmbeddings |
Raw image patches linearly embedded (e.g., ViLT, ViT-based). |
VisualFeatureType RegionFeatures |
Object-level features from a detection model (e.g., Faster R-CNN). |
int value__ |