.png){kind=icon}
.png){kind=icon}
Return
Class-specific image segmentation across multiple domains using customized U-Net pipelines
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2025.129203.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A modular training pipeline using class- and domain-specific U-Net models. • Using an F1-based loss function to boost multi-image segmentation performance. • Enhanced preprocessing and postprocessing techniques for robustness. • Comprehensive evaluation across multiple visual domains with real-world data.
Keywords:
U-Net
multi-image segmentation
F1 loss
preprocessing
postprocessing
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W