Fig 1.
The architecture of OD-MVSNet.
Fig 2.
The omni-dimensional dynamic atrous spatial pyramid pooling module.
Fig 3.
The atrous convolution.
Fig 4.
The omni-dimensional dynamic convolution module.
Fig 5.
Normalization-based 3D attention module.
Fig 6.
The problem of similarity confidence at different depth.
Fig 7.
Qualitative results of scan9 of the DTU dataset.
(a) MVSNet, (b) CasMVSNet, (c) Ours.
Table 1.
Quantitative results on the DTU dataset (lower is better).
Table 2.
Comparison of GPU memory consumption and inference time on the DTU dataset.
Fig 8.
Comparison of point cloud reconstructed images from different modules.
Baseline, Baseline+OSPP, Baseline+N3DAM, OD-MVSNet.
Table 3.
Ablation studies on the DTU dataset.