MVSFormer: Multi-View Stereo by Learning Robust Image Features and Temperature-based Depth
Chenjie Cao, Xinlin Ren, Yanwei Fu
OpenReviewMVSFormer: Multi-View Stereo by Learning Robust Image Features and...Feature representation learning is the key recipe for learning-based Multi-View Stereo (MVS). As the common feature extractor of learning-based MVS, vanilla Feature Pyramid Networks (FPNs) suffer...