Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chen, Rui, Zhang, Jianfeng, Liang, Yixun, Luo, Guan, Li, Weiyu, Liu, Jiarui, Li, Xiu, Long, Xiaoxiao, Feng, Jiashi, Tan, Ping
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915211822235648
author Chen, Rui
Zhang, Jianfeng
Liang, Yixun
Luo, Guan
Li, Weiyu
Liu, Jiarui
Li, Xiu
Long, Xiaoxiao
Feng, Jiashi
Tan, Ping
author_facet Chen, Rui
Zhang, Jianfeng
Liang, Yixun
Luo, Guan
Li, Weiyu
Liu, Jiarui
Li, Xiu
Long, Xiaoxiao
Feng, Jiashi
Tan, Ping
contents Recent 3D content generation pipelines commonly employ Variational Autoencoders (VAEs) to encode shapes into compact latent representations for diffusion-based generation. However, the widely adopted uniform point sampling strategy in Shape VAE training often leads to a significant loss of geometric details, limiting the quality of shape reconstruction and downstream generation tasks. We present Dora-VAE, a novel approach that enhances VAE reconstruction through our proposed sharp edge sampling strategy and a dual cross-attention mechanism. By identifying and prioritizing regions with high geometric complexity during training, our method significantly improves the preservation of fine-grained shape features. Such sampling strategy and the dual attention mechanism enable the VAE to focus on crucial geometric details that are typically missed by uniform sampling approaches. To systematically evaluate VAE reconstruction quality, we additionally propose Dora-bench, a benchmark that quantifies shape complexity through the density of sharp edges, introducing a new metric focused on reconstruction accuracy at these salient geometric features. Extensive experiments on the Dora-bench demonstrate that Dora-VAE achieves comparable reconstruction quality to the state-of-the-art dense XCube-VAE while requiring a latent space at least 8$\times$ smaller (1,280 vs. > 10,000 codes).
format Preprint
id arxiv_https___arxiv_org_abs_2412_17808
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders
Chen, Rui
Zhang, Jianfeng
Liang, Yixun
Luo, Guan
Li, Weiyu
Liu, Jiarui
Li, Xiu
Long, Xiaoxiao
Feng, Jiashi
Tan, Ping
Computer Vision and Pattern Recognition
Recent 3D content generation pipelines commonly employ Variational Autoencoders (VAEs) to encode shapes into compact latent representations for diffusion-based generation. However, the widely adopted uniform point sampling strategy in Shape VAE training often leads to a significant loss of geometric details, limiting the quality of shape reconstruction and downstream generation tasks. We present Dora-VAE, a novel approach that enhances VAE reconstruction through our proposed sharp edge sampling strategy and a dual cross-attention mechanism. By identifying and prioritizing regions with high geometric complexity during training, our method significantly improves the preservation of fine-grained shape features. Such sampling strategy and the dual attention mechanism enable the VAE to focus on crucial geometric details that are typically missed by uniform sampling approaches. To systematically evaluate VAE reconstruction quality, we additionally propose Dora-bench, a benchmark that quantifies shape complexity through the density of sharp edges, introducing a new metric focused on reconstruction accuracy at these salient geometric features. Extensive experiments on the Dora-bench demonstrate that Dora-VAE achieves comparable reconstruction quality to the state-of-the-art dense XCube-VAE while requiring a latent space at least 8$\times$ smaller (1,280 vs. > 10,000 codes).
title Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.17808