CObL: Toward Zero-Shot Ordinal Layering without User Prompting

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Damaraju, Aneel, Hazineh, Dean, Zickler, Todd
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918122777214976
author Damaraju, Aneel
Hazineh, Dean
Zickler, Todd
author_facet Damaraju, Aneel
Hazineh, Dean
Zickler, Todd
contents Vision benefits from grouping pixels into objects and understanding their spatial relationships, both laterally and in depth. We capture this with a scene representation comprising an occlusion-ordered stack of "object layers," each containing an isolated and amodally-completed object. To infer this representation from an image, we introduce a diffusion-based architecture named Concurrent Object Layers (CObL). CObL generates a stack of object layers in parallel, using Stable Diffusion as a prior for natural objects and inference-time guidance to ensure the inferred layers composite back to the input image. We train CObL using a few thousand synthetically-generated images of multi-object tabletop scenes, and we find that it zero-shot generalizes to photographs of real-world tabletops with varying numbers of novel objects. In contrast to recent models for amodal object completion, CObL reconstructs multiple occluded objects without user prompting and without knowing the number of objects beforehand. Unlike previous models for unsupervised object-centric representation learning, CObL is not limited to the world it was trained in.
format Preprint
id arxiv_https___arxiv_org_abs_2508_08498
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CObL: Toward Zero-Shot Ordinal Layering without User Prompting
Damaraju, Aneel
Hazineh, Dean
Zickler, Todd
Computer Vision and Pattern Recognition
Vision benefits from grouping pixels into objects and understanding their spatial relationships, both laterally and in depth. We capture this with a scene representation comprising an occlusion-ordered stack of "object layers," each containing an isolated and amodally-completed object. To infer this representation from an image, we introduce a diffusion-based architecture named Concurrent Object Layers (CObL). CObL generates a stack of object layers in parallel, using Stable Diffusion as a prior for natural objects and inference-time guidance to ensure the inferred layers composite back to the input image. We train CObL using a few thousand synthetically-generated images of multi-object tabletop scenes, and we find that it zero-shot generalizes to photographs of real-world tabletops with varying numbers of novel objects. In contrast to recent models for amodal object completion, CObL reconstructs multiple occluded objects without user prompting and without knowing the number of objects beforehand. Unlike previous models for unsupervised object-centric representation learning, CObL is not limited to the world it was trained in.
title CObL: Toward Zero-Shot Ordinal Layering without User Prompting
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.08498