BriefGPT.xyz
Jun, 2021
LegoFormer：面块级别多视角三维重构的变形器
LegoFormer: Transformers for Block-by-Block Multi-view 3D Reconstruction
HTML
PDF
Farid Yagubbayli, Alessio Tonioni, Federico Tombari
TL;DR
LegoFormer是一种基于transformer模型的体素3D重建技术，通过使用自注意力层在所有计算阶段分享视角之间的信息，并将输出分解为低秩矩阵，从而实现对各个独立结构的预测和聚合，具有竞争性的性能和可解释性的优点，可用于现实数据的广义重建任务。
Abstract
Most modern
deep learning
-based
multi-view 3d reconstruction
techniques use RNNs or fusion modules to combine information from multiple images after encoding them. These two separate steps have loose connections
→