Skip to main navigation Skip to search Skip to main content

Multi-Hypothesis 3D Hand Mesh Recovering from a Single Blurry Image

  • Yuming Chen
  • , Rongyu Chen
  • , Zhongqun Zhang
  • , Yihua Cheng*
  • , Hyung Jin Chang
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

84 Downloads (Pure)

Abstract

Recovery of 3D hand mesh from blurry hand images is challenging due to the ambiguity. Most existing works attempt to solve this issue by exploiting physical and temporal constraints. However, those works ignore the fact that multiple feasible solutions exist. In this paper, we propose a two-stage Multi-Hypothesis Hand Mesh Recovery network, consisting of a generation and selection model. In the first stage, the generation model explicitly extracts the temporal information with an unfolder. Then, a multi-hypothesis Transformer generates multiple diverse hypotheses with a lightweight hypothesis embedding set. In the second stage, the selection model selects a subset of good-quality hypotheses. We additionally combine the classifying and ranking loss to better align with the target of the selection model. Extensive experiments show that the proposed method produces much more accurate results on blurry images. Source code is available at https://github.com/RandSF/Multi_Hypothesis_BlurHandNet.
Original languageEnglish
Title of host publication2025 IEEE International Conference on Multimedia and Expo (ICME)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Number of pages8
ISBN (Electronic)9798331594954
ISBN (Print)9798331594961 (PoD)
DOIs
Publication statusPublished - 30 Oct 2025
Event2025 IEEE International Conference on Multimedia and Expo (ICME) - Nantes, France
Duration: 30 Jun 2025 → 4 Jul 2025

Publication series

NameIEEE International Conference on Multimedia and Expo
PublisherIEEE
ISSN (Print)1945-7871
ISSN (Electronic)1945-788X

Conference

Conference2025 IEEE International Conference on Multimedia and Expo (ICME)
Period30/06/25 → 4/07/25

Keywords

  • Hands
  • Computer vision
  • Three-dimensional displays
  • Accuracy
  • Source coding
  • Computational modeling
  • Reinforcement learning
  • Transformers
  • Reliability
  • Data mining

Fingerprint

Dive into the research topics of 'Multi-Hypothesis 3D Hand Mesh Recovering from a Single Blurry Image'. Together they form a unique fingerprint.

Cite this