Authors: Xiaozheng Zheng, Chao Wen, Zhuo Su, Zeran Xu, Zhaohu Li, Yang Zhao, Zhou Xue

Published on: February 29, 2024

Impact Score: 7.6

Arxiv code: Arxiv:2402.18969

Summary

• What is new: A novel method for creating high-fidelity, animated hand avatars from just one image, leveraging a hand prior model.
• Why this is important: The need for quick and personalized hand avatar creation in the digital human domain, with existing methods requiring extensive data and being impractical.
• What the research proposes: OHTA (One-shot Hand avaTAr), a new method that creates detailed hand avatars from a single image by learning and using data-driven hand priors.
• Results: Successfully generated high-fidelity and animatable hand avatars from one image, demonstrating versatility in text-to-avatar conversion, hand editing, and latent space manipulation.

Technical Details

Technological frameworks used: nan

Models used: Hand prior model for learning data-driven priors and inversion fitting.

Data used: nan

Potential Impact

VR/AR, gaming, virtual meeting platforms, animation, and digital content creation sectors.

Want to implement this idea in a business?

We have generated a startup concept here: HandAI.