Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• 4DEquine disentangles 4D equine reconstruction into dynamic motion reconstruction and static appearance reconstruction.method 95%
• A spatio-temporal transformer with a post-optimization stage is used for dynamic motion reconstruction.method 95%
• A feed-forward network reconstructs a high-fidelity, animatable 3D Gaussian avatar from a single image for appearance reconstruction.method 95%
• The VarenPoser dataset is a large-scale synthetic motion dataset with high-quality surface motions and diverse camera trajectories.technical 95%
• The VarenTex dataset is a synthetic appearance dataset comprising realistic multi-view images generated through multi-view diffusion.technical 95%
• 4DEquine achieves state-of-the-art performance on real-world APT36K and AiM datasets for both geometry and appearance reconstruction, despite training only on synthetic datasets.result 95%
• Previous mainstream 4D animal reconstruction methods require joint optimization of motion and appearance over a whole video, which is time-consuming and sensitive to incomplete observation.limitation 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card