MaskDiceLoss<T>

Models & Types · AiDotNet.ComputerVision.Segmentation.Losses

Dice loss for mask prediction.

For Beginners

Dice loss directly optimizes the Dice coefficient (F1 score), making it better for imbalanced masks where foreground is much smaller than background.

Constructors

Constructor Summary
MaskDiceLoss(double smooth) Creates a new mask Dice loss.

Methods

Method Summary
T Compute(Tensor<T> predicted, Tensor<T> target) Computes Dice loss between predicted and target masks.