LiteMatch uses CVCE and HFE encoders plus CVC-Loss to enable lightweight zero-shot stereo matching with competitive EPE and D1 scores on Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo using 3.36M-9.58M parameters.
Monster++: Unified stereo matching, multi- view stereo, and real-time stereo with monodepth priors
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CV 3years
2026 3representative citing papers
GREATEN fuses surface normals into stereo matching with gated contextual-geometric fusion and sparse attention, cutting Syn-to-Real errors by up to 30% when trained only on synthetic data.
LAS2 is a series of efficient stereo matching models that reach state-of-the-art zero-shot performance among fast methods while running 1.8-2.7x faster than prior iterative approaches on H200 and Orin hardware.
citing papers explorer
-
LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization
LiteMatch uses CVCE and HFE encoders plus CVC-Loss to enable lightweight zero-shot stereo matching with competitive EPE and D1 scores on Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo using 3.36M-9.58M parameters.
-
Geometry Reinforced Efficient Attention Tuning Equipped with Normals for Robust Stereo Matching
GREATEN fuses surface normals into stereo matching with gated contextual-geometric fusion and sparse attention, cutting Syn-to-Real errors by up to 30% when trained only on synthetic data.
-
Lite Any Stereo V2: Faster and Stronger Efficient Zero-Shot Stereo Matching
LAS2 is a series of efficient stereo matching models that reach state-of-the-art zero-shot performance among fast methods while running 1.8-2.7x faster than prior iterative approaches on H200 and Orin hardware.