N Heads Are Better than One: Exploring Theoretical Performance Bounds of 3D Face Reconstruction Methods

Abstract

We introduce “N Heads Are Better Than One”, a novel approach for evaluating combinations of existing 3D face reconstruction methods. By calculating lower theoretical error bounds for method combinations on the NoW benchmark, we establish a robust set of new baselines for the task of 3D face reconstruction. Our work also provides a framework for assessing the potential of these aggregate ‘pseudo-foundation models,’ which leverage strengths from multiple existing approaches. In doing so, we improve understanding of the performance of current methods and set targets for future foundation models to beat.

Cite

Text

Rowan et al. "N Heads Are Better than One: Exploring Theoretical Performance Bounds of 3D Face Reconstruction Methods." European Conference on Computer Vision Workshops, 2024. doi:10.1007/978-3-031-92591-7_28

Markdown

[Rowan et al. "N Heads Are Better than One: Exploring Theoretical Performance Bounds of 3D Face Reconstruction Methods." European Conference on Computer Vision Workshops, 2024.](https://mlanthology.org/eccvw/2024/rowan2024eccvw-heads/) doi:10.1007/978-3-031-92591-7_28

BibTeX

@inproceedings{rowan2024eccvw-heads,
  title     = {{N Heads Are Better than One: Exploring Theoretical Performance Bounds of 3D Face Reconstruction Methods}},
  author    = {Rowan, Will and Huber, Patrik and Pears, Nick E. and Keeling, Andrew},
  booktitle = {European Conference on Computer Vision Workshops},
  year      = {2024},
  pages     = {427-435},
  doi       = {10.1007/978-3-031-92591-7_28},
  url       = {https://mlanthology.org/eccvw/2024/rowan2024eccvw-heads/}
}