Reproduce “TRUST-Stereo: Teacher-Assisted Robust Uncertainty-Guided Search and Transfer for Fast Stereo Matching” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pouya Sohrabipour, Yihong Feng, Chaitanya Kumar Reddy Pallerla, Tuan Dang, Ngan Le, Dongyi Wang
TRUST-Stereo predicts when propagated disparity estimates may miss the next local search window, then shifts that window’s center without widening it. At inference, it retains the candidate count, while a teacher-assistant strategy strengthens lightweight matching features. In SceneFlow-only zero-shot evaluation, it achieves the best result on five of six metrics; ETH3D Bad 1.0 is the exception.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.