Summary of Fine Structure-aware Sampling: a New Sampling Training Scheme For Pixel-aligned Implicit Models in Single-view Human Reconstruction, by Kennard Yanting Chan et al.
Fine Structure-Aware Sampling: A New Sampling Training Scheme for Pixel-Aligned Implicit Models in Single-View Human Reconstruction
by Kennard Yanting Chan, Fayao Liu, Guosheng Lin, Chuan Sheng Foo, Weisi Lin
First submitted to arxiv on: 29 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces Fine Structured-Aware Sampling (FSS), a new training scheme for pixel-aligned implicit models, specifically designed for single-view clothed human reconstruction. FSS addresses the limitations of existing sampling schemes by adaptively capturing thin surfaces and reducing noisy artefacts in reconstructed meshes. Unlike previous methods, FSS utilizes normals of sample points to improve results and proposes a mesh thickness loss signal to enhance the training process. By reworking the pixel-aligned implicit function framework, FSS becomes computationally feasible. The proposed method significantly outperforms state-of-the-art (SOTA) methods both qualitatively and quantitatively. PIFu, PIFuHD, ICON, and SOTA methods are mentioned as relevant to this research. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about improving how we use computers to create 3D models of people from just one photo. Right now, these computers have trouble capturing thin parts like ears or fingers, and they can make mistakes that look like noise. The researchers created a new way to train these computers called Fine Structured-Aware Sampling (FSS). FSS helps the computers better capture thin surfaces and reduces errors. They also found a way to use the direction of surface normal points to make the results even better. This new method is very good at making 3D models, and it’s available for others to use. |
