用曲率正则化提升稀疏深度下的3D场景重建精度
Curvature-Regularized Variational Autoencoder for 3D Scene Reconstruction from Sparse Depth
- 引入离散拉普拉斯算子实现曲率正则化
- 在仅5%深度数据下比标准VAE高18.1%精度
- 单个正则项优于复杂多约束组合，训练开销仅15%
当深度传感器仅提供所需测量值的5%时，完整重建3D场景变得困难。自动驾驶车辆和机器人无法容忍稀疏重建带来的几何误差。我们通过离散拉普拉斯算子引入曲率正则化，使重建精度比标准变分自编码器提升18.1%。本工作挑战了几何深度学习中的一个隐含假设：多个几何约束的组合必然提升性能。结果表明，单一精心设计的正则化项不仅达到，甚至超越了复杂多术语公式的有效性。离散拉普拉斯算子在仅增加15%训练开销的情况下，提供稳定梯度与噪声抑制，且推理无额外成本。代码与模型已公开于 https://github.com/Maryousefi/GeoVAE-3D。
原文摘要 · Abstract (English)
When depth sensors provide only 5% of needed measurements, reconstructing complete 3D scenes becomes difficult. Autonomous vehicles and robots cannot tolerate the geometric errors that sparse reconstruction introduces. We propose curvature regularization through a discrete Laplacian operator, achieving 18.1% better reconstruction accuracy than standard variational autoencoders. Our contribution challenges an implicit assumption in geometric deep learning: that combining multiple geometric constraints improves performance. A single well-designed regularization term not only matches but exceeds the effectiveness of complex multi-term formulations. The discrete Laplacian offers stable gradients and noise suppression with just 15% training overhead and zero inference cost. Code and models are available at https://github.com/Maryousefi/GeoVAE-3D.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。