NeRD++: Improved 3D-mirror symmetry learning from a single image
Yancong Lin ( Delft University of Technology),* Silvia-Laura L Pintea (TU Delft), Jan C van Gemert (Delft University of Technology)
The 33rd British Machine Vision Conference
Abstract
Many objects are naturally symmetric, and this symmetry can be exploited to infer unseen 3D properties from a single 2D image. Recently, NeRD is proposed for accurate 3D mirror plane estimation from a single image. Despite the unprecedented accuracy, it relies on large annotated datasets for training and suffers from slow inference. Here we aim to improve its data and compute efficiency. We do away with the computationally expensive 4D feature volumes and instead explicitly compute the feature correlation of the pixel correspondences across depth, thus creating a compact 3D volume. We also design multi-stage spherical convolutions to identify the optimal mirror plane on the hemisphere, whose inductive bias offers gains in data-efficiency. Experiments on both synthetic and real-world datasets show the benefit of our proposed changes for improved data efficiency and inference speed.Video
Citation
@inproceedings{Lin_2022_BMVC,
author = {Yancong Lin and Silvia-Laura L Pintea and Jan C van Gemert},
title = {NeRD++: Improved 3D-mirror symmetry learning from a single image},
booktitle = {33rd British Machine Vision Conference 2022, {BMVC} 2022, London, UK, November 21-24, 2022},
publisher = {{BMVA} Press},
year = {2022},
url = {https://bmvc2022.mpi-inf.mpg.de/0223.pdf}
}
Copyright © 2022
The British Machine Vision Conference is organised by