RIFT models achieve top or tied performance on four crack segmentation benchmarks with a lightweight morphology-aligned design, including a 0.47M-parameter variant.
InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 16560–16569
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Rethinking Efficient Crack Segmentation with Task-Aligned Structural-Directional Modeling
RIFT models achieve top or tied performance on four crack segmentation benchmarks with a lightweight morphology-aligned design, including a 0.47M-parameter variant.