DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humans
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916186409664512 |
|---|---|
| author | Sengupta, Akash Alldieck, Thiemo Kolotouros, Nikos Corona, Enric Zanfir, Andrei Sminchisescu, Cristian |
| author_facet | Sengupta, Akash Alldieck, Thiemo Kolotouros, Nikos Corona, Enric Zanfir, Andrei Sminchisescu, Cristian |
| contents | We present DiffHuman, a probabilistic method for photorealistic 3D human reconstruction from a single RGB image. Despite the ill-posed nature of this problem, most methods are deterministic and output a single solution, often resulting in a lack of geometric detail and blurriness in unseen or uncertain regions. In contrast, DiffHuman predicts a probability distribution over 3D reconstructions conditioned on an input 2D image, which allows us to sample multiple detailed 3D avatars that are consistent with the image. DiffHuman is implemented as a conditional diffusion model that denoises pixel-aligned 2D observations of an underlying 3D shape representation. During inference, we may sample 3D avatars by iteratively denoising 2D renders of the predicted 3D representation. Furthermore, we introduce a generator neural network that approximates rendering with considerably reduced runtime (55x speed up), resulting in a novel dual-branch diffusion framework. Our experiments show that DiffHuman can produce diverse and detailed reconstructions for the parts of the person that are unseen or uncertain in the input image, while remaining competitive with the state-of-the-art when reconstructing visible surfaces. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_00485 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humans Sengupta, Akash Alldieck, Thiemo Kolotouros, Nikos Corona, Enric Zanfir, Andrei Sminchisescu, Cristian Computer Vision and Pattern Recognition We present DiffHuman, a probabilistic method for photorealistic 3D human reconstruction from a single RGB image. Despite the ill-posed nature of this problem, most methods are deterministic and output a single solution, often resulting in a lack of geometric detail and blurriness in unseen or uncertain regions. In contrast, DiffHuman predicts a probability distribution over 3D reconstructions conditioned on an input 2D image, which allows us to sample multiple detailed 3D avatars that are consistent with the image. DiffHuman is implemented as a conditional diffusion model that denoises pixel-aligned 2D observations of an underlying 3D shape representation. During inference, we may sample 3D avatars by iteratively denoising 2D renders of the predicted 3D representation. Furthermore, we introduce a generator neural network that approximates rendering with considerably reduced runtime (55x speed up), resulting in a novel dual-branch diffusion framework. Our experiments show that DiffHuman can produce diverse and detailed reconstructions for the parts of the person that are unseen or uncertain in the input image, while remaining competitive with the state-of-the-art when reconstructing visible surfaces. |
| title | DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humans |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2404.00485 |