Instance Normalization (also known as contrast normalization) is a normalization layer where:
$$ y_{tijk} = \frac{x_{tijk} - \mu_{ti}}{\sqrt{\sigma_{ti}^2 + \epsilon}}, \quad \mu_{ti} = \frac{1}{HW}\sum_{l=1}^W \sum_{m=1}^H x_{tilm}, \quad \sigma_{ti}^2 = \frac{1}{HW}\sum_{l=1}^W \sum_{m=1}^H (x_{tilm} - \mu_{ti})^2. $$
This prevents instance-specific mean and covariance shift simplifying the learning process. Intuitively, the normalization process allows to remove instance-specific contrast information from the content image in a task like image stylization, which simplifies generation.
Source: Instance Normalization: The Missing Ingredient for Fast StylizationPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Image-to-Image Translation | 82 | 10.41% |
Domain Adaptation | 46 | 5.84% |
Style Transfer | 44 | 5.58% |
Image Generation | 44 | 5.58% |
Semantic Segmentation | 40 | 5.08% |
Object Detection | 20 | 2.54% |
Unsupervised Domain Adaptation | 20 | 2.54% |
Image Segmentation | 19 | 2.41% |
Domain Generalization | 14 | 1.78% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |