CA-I2P combines channel-adaptive feature adjustment and optimal-transport-based global selection to improve image-to-point cloud registration, reporting SOTA recall on RGB-D Scenes V2 (63.3% RR) and 7-Scenes (79.5% RR).
Revisiting domain gener- alized stereo matching networks from a feature consistency perspective
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CA-I2P: Channel-Adaptive Registration Network with Global Optimal Selection
CA-I2P combines channel-adaptive feature adjustment and optimal-transport-based global selection to improve image-to-point cloud registration, reporting SOTA recall on RGB-D Scenes V2 (63.3% RR) and 7-Scenes (79.5% RR).