S²VAE replaces Gaussian bottlenecks with hyperspherical Power Spherical latents in a VAE on VGGT features, yielding better results on depth estimation, camera pose recovery, and point cloud reconstruction especially at high compression.
To do this, the CLIP model has two separate encoders (one for images, one for text)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond Gaussian Bottlenecks: Topologically Aligned Encoding of Vision-Transformer Feature Spaces
S²VAE replaces Gaussian bottlenecks with hyperspherical Power Spherical latents in a VAE on VGGT features, yielding better results on depth estimation, camera pose recovery, and point cloud reconstruction especially at high compression.