ODISEModelSize
Enums · AiDotNet.Enums
Defines the backbone size variants for ODISE (Open-vocabulary DIffusion-based panoptic SEgmentation).
For Beginners
ODISE uses text-to-image diffusion features fused with discriminative features for open-vocabulary panoptic segmentation.
How It Works
Reference: Xu et al., “Open-Vocabulary Panoptic Segmentation with Text-to-Image Diffusion Models”, CVPR 2023 Highlight.
Fields
| Field | Summary |
|---|---|
ODISEModelSize Base |
Base variant with Stable Diffusion backbone. |
ODISEModelSize Large |
Large variant with larger CLIP and diffusion models. |
int value__ |