Summary of Regennet: Towards Human Action-reaction Synthesis, by Liang Xu et al.
ReGenNet: Towards Human Action-Reaction Synthesis
by Liang Xu, Yizhou Zhou, Yichao Yan, Xin Jin, Wenhan Zhu, Fengyun Rao, Xiaokang Yang, Wenjun Zeng
First submitted to arxiv on: 18 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
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 The paper proposes a novel approach for synthesizing dynamic human-human interactions by analyzing the asymmetric nature of such interactions. The authors develop a multi-setting benchmark for generating human reactions conditioned on given actions, using datasets like NTU120, InterHuman, and Chi3D. They propose a diffusion-based generative model called ReGenNet, with a Transformer decoder architecture, to predict human reactions online. The method incorporates an explicit distance-based interaction loss to account for the dynamic nature of interactions. Results show that the proposed approach outperforms baselines in generating plausible and instantaneous human reactions, and generalizes well to unseen scenarios. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Humans interact with each other constantly, but most AI models focus on static scenes or objects. This paper wants to change that by understanding how humans react to each other’s actions. They analyze how people interact and propose a way to generate those reactions using a special kind of AI model called ReGenNet. The team tested their approach using several datasets and found it can create realistic human reactions in different scenarios. |
Keywords
» Artificial intelligence » Decoder » Diffusion » Generative model » Transformer
