New InterPet4D dataset and InterPetMoGen framework advance human-pet motion generation

Researchers have introduced InterPet4D, a novel multimodal dataset designed to advance the study of human-pet interactions, specifically focusing on motion generation. This dataset comprises 6.8 million frames capturing natural interactions between humans and dogs, featuring synchronized multi-view videos, 2D/3D keypoints, meshes, and audio. Alongside the dataset, the team developed the InterPetMoGen framework, which utilizes a Diffusion Transformer architecture. Their model demonstrated superior performance over baseline methods, achieving an FID score of 11.21, highlighting the dataset's utility for realistic human-pet motion modeling. AI

IMPACT This dataset and framework could enable new research and applications in realistic human-pet interaction modeling and animation.

RANK_REASON The cluster describes a new academic dataset and framework for motion generation, published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Diffusion Transformer
• Gotit.pub
• Hugging Face
• InterPet4D
• InterPetMoGen
• ScienceCast
• sequence-to-sequence learning

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →