[논문리뷰] TransNormal-2: Geometry-Grounded Rectified Flow with Edge-Aware Decoding for Precise Normal Estimation본 논문은 diffusion-based monocular geometry estimation에서 간과되었던 VAE reconstruction degradation이라는 근본적인 문제점을 제기한다.#Review#Surface Normal Estimation#Monocular Geometry Estimation#Latent Diffusion Models#Transparent Object Perception#VAE Reconstruction Degradation#Rectified Flow#Geometric Refinement Module2026년 9월 8일댓글 수 로딩 중