Composing per-concept diffusion models and inverting them with denoising loss enables multi-object scene understanding that generalizes beyond the training distribution.
Towards compositional understanding of the world by agent-based deep learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Compositional Scene Understanding through Inverse Generative Modeling
Composing per-concept diffusion models and inverting them with denoising loss enables multi-object scene understanding that generalizes beyond the training distribution.