DatasetNeRF: Efficient 3D-aware Data Factory with Generative Radiance Fields
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911991781654528 |
|---|---|
| author | Chi, Yu Zhan, Fangneng Wu, Sibo Theobalt, Christian Kortylewski, Adam |
| author_facet | Chi, Yu Zhan, Fangneng Wu, Sibo Theobalt, Christian Kortylewski, Adam |
| contents | Progress in 3D computer vision tasks demands a huge amount of data, yet annotating multi-view images with 3D-consistent annotations, or point clouds with part segmentation is both time-consuming and challenging. This paper introduces DatasetNeRF, a novel approach capable of generating infinite, high-quality 3D-consistent 2D annotations alongside 3D point cloud segmentations, while utilizing minimal 2D human-labeled annotations. Specifically, we leverage the strong semantic prior within a 3D generative model to train a semantic decoder, requiring only a handful of fine-grained labeled samples. Once trained, the decoder efficiently generalizes across the latent space, enabling the generation of infinite data. The generated data is applicable across various computer vision tasks, including video segmentation and 3D point cloud segmentation. Our approach not only surpasses baseline models in segmentation quality, achieving superior 3D consistency and segmentation precision on individual images, but also demonstrates versatility by being applicable to both articulated and non-articulated generative models. Furthermore, we explore applications stemming from our approach, such as 3D-aware semantic editing and 3D inversion. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2311_12063 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | DatasetNeRF: Efficient 3D-aware Data Factory with Generative Radiance Fields Chi, Yu Zhan, Fangneng Wu, Sibo Theobalt, Christian Kortylewski, Adam Computer Vision and Pattern Recognition Progress in 3D computer vision tasks demands a huge amount of data, yet annotating multi-view images with 3D-consistent annotations, or point clouds with part segmentation is both time-consuming and challenging. This paper introduces DatasetNeRF, a novel approach capable of generating infinite, high-quality 3D-consistent 2D annotations alongside 3D point cloud segmentations, while utilizing minimal 2D human-labeled annotations. Specifically, we leverage the strong semantic prior within a 3D generative model to train a semantic decoder, requiring only a handful of fine-grained labeled samples. Once trained, the decoder efficiently generalizes across the latent space, enabling the generation of infinite data. The generated data is applicable across various computer vision tasks, including video segmentation and 3D point cloud segmentation. Our approach not only surpasses baseline models in segmentation quality, achieving superior 3D consistency and segmentation precision on individual images, but also demonstrates versatility by being applicable to both articulated and non-articulated generative models. Furthermore, we explore applications stemming from our approach, such as 3D-aware semantic editing and 3D inversion. |
| title | DatasetNeRF: Efficient 3D-aware Data Factory with Generative Radiance Fields |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2311.12063 |