Skip to main content
. 2024 Mar 16;24(6):1919. doi: 10.3390/s24061919

Table 3.

Quantitative comparison of two representative scenes from ScanNet and 3Dmatch.

3DMatch 7-Scenes-Fire ScanNet Scene0002_00
PSNR↑ SSIM↑ LPIPS↓ Depth-RMSE↓ PSNR↑ SSIM↑ LPIPS↓ Depth-RMSE↓
Depth-NeRF w/o completion 29.78 0.549 0.038 0.173 28.55 0.496 0.278 0.204
Depth-NeRF w/o depth loss 30.31 0.572 0.099 1.115 29.59 0.534 0.320 0.524
Depth-NeRF w/o dynamic sampling 30.40 0.564 0.128 0.498 29.48 0.531 0.231 0.489
Depth-NeRF 30.41 0.592 0.026 0.152 30.03 0.566 0.186 0.201