ViTCoMerModelSize

Enums · AiDotNet.Enums

Defines the model size variants for ViT-CoMer (hybrid CNN-Transformer).

For Beginners

ViT-CoMer combines a CNN branch with a Vision Transformer branch to get the best of both worlds — CNNs excel at fine local details while transformers capture global context. The result is better boundary quality in segmentation.

How It Works

Technical Details: Each size uses a different ViT backbone paired with a CNN branch of matching capacity. The two branches exchange information through multi-scale feature interaction modules.

Reference: Xia et al., “ViT-CoMer: Vision Transformer with Convolutional Multi-scale Feature Interaction for Dense Predictions”, CVPR 2024.

Fields

Field Summary
ViTCoMerModelSize Base ViT-CoMer-B: Base variant (100M params).
ViTCoMerModelSize Large ViT-CoMer-L: Large variant (350M params).
ViTCoMerModelSize Small ViT-CoMer-S: Small variant (45M params).
int value__