InterEdit: Navigating Text-Guided Multi-Human 3D Motion Editing

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• InterEdit introduces a new dataset called InterEdit3D with manual two-person motion change annotations to support multi-person 3D motion editing.method 95%
  Evidencepartial

  We propose InterEdit3D, a new dataset with manual two-person motion change annotations

  Implicationpartial

  Explicitly stated in the abstract as a core contribution of the paper

  Verificationpartial

  partial

• InterEdit introduces Semantic-Aware Plan Token Alignment with learnable tokens to capture high-level interaction cues.technical 90%
  Evidencepartial

  It introduces Semantic-Aware Plan Token Alignment with learnable tokens to capture high-level interaction cues

  Implicationpartial

  Directly stated in the abstract as a specific technical component of the method

  Verificationpartial

  partial

• InterEdit uses an Interaction-Aware Frequency Token Alignment strategy with DCT and energy pooling to model periodic motion dynamics.technical 90%
  Evidencepartial

  an Interaction-Aware Frequency Token Alignment strategy using DCT and energy pooling to model periodic motion dynamics

  Implicationpartial

  Directly stated in the abstract as a specific technical component of the method

  Verificationpartial

  partial

• InterEdit is a synchronized classifier-free conditional diffusion model for Text-guided Multi-human Motion Editing (TMME).method 90%
  Evidencepartial

  We present InterEdit, a synchronized classifier-free conditional diffusion model for TMME

  Implicationpartial

  Directly stated in the abstract describing the model architecture

  Verificationpartial

  partial

• InterEdit improves text-to-motion consistency and edit fidelity compared to existing methods.result 85%
  Evidencepartial

  Experiments show that InterEdit improves text-to-motion consistency and edit fidelity

  Implicationpartial

  Directly stated in the abstract as an experimental result, though specific metrics are not provided

  Verificationpartial

  partial

• InterEdit achieves state-of-the-art performance on the Text-guided Multi-human Motion Editing (TMME) benchmark.result 85%
  Evidencepartial

  achieving state-of-the-art TMME performance

  Implicationpartial

  Directly stated in the abstract as an experimental result, though specific metrics are not provided

  Verificationpartial

  partial

• Text-guided 3D motion editing has seen success in single-person scenarios but is less explored in multi-person settings due to limited paired data and complex inter-person interactions.limitation 80%
  Evidencepartial

  Text-guided 3D motion editing has seen success in single-person scenarios, but its extension to multi-person settings is less explored due to limited paired data and the complexity of inter-person interactions

  Implicationpartial

  Directly stated in the abstract as motivation for the work, though it describes the field rather than the paper's specific contribution

  Verificationpartial

  partial

• The dataset and code for InterEdit will be publicly released at a specific GitHub repository.method 95%
  Evidencepartial

  The dataset and code will be released at https://github.com/YNG916/InterEdit

  Implicationpartial

  Explicitly stated in the abstract with a specific URL provided

  Verificationpartial

  partial

Startup potential card