Metrics and Losses

Losses

• Catagorical Crossentropy
• Binary Crossentropy
• Crossentropy + SSIM
• Dice
• Crossentropy + Dice
• Tversky
• Focal
• Focal + Tversky

Code

from tf_semantic_segmentation.losses import get_loss_by_name, losses_by_name

losses = list(losses_by_name.keys())
for l in losses:
    fn = get_loss_by_name(l)
    value = fn(y_true, y_pred)

Metrics

• f1
• f2
• iou
• precision
• recall
• psnr
• ssim

Using Code

from tf_semantic_segmentation.metrics import get_metric_by_name, metrics_by_name

metrics = list(metrics_by_name.keys())
for m in metrics:
    fn = get_metric_by_name(m)
    value = fn(y_true, y_pred)