LatentCRF: Continuous CRF for Efficient Latent Diffusion

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ranasinghe, Kanchana, Jayasumana, Sadeep, Veit, Andreas, Chakrabarti, Ayan, Glasner, Daniel, Ryoo, Michael S, Ramalingam, Srikumar, Kumar, Sanjiv
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915078671958016
author Ranasinghe, Kanchana
Jayasumana, Sadeep
Veit, Andreas
Chakrabarti, Ayan
Glasner, Daniel
Ryoo, Michael S
Ramalingam, Srikumar
Kumar, Sanjiv
author_facet Ranasinghe, Kanchana
Jayasumana, Sadeep
Veit, Andreas
Chakrabarti, Ayan
Glasner, Daniel
Ryoo, Michael S
Ramalingam, Srikumar
Kumar, Sanjiv
contents Latent Diffusion Models (LDMs) produce high-quality, photo-realistic images, however, the latency incurred by multiple costly inference iterations can restrict their applicability. We introduce LatentCRF, a continuous Conditional Random Field (CRF) model, implemented as a neural network layer, that models the spatial and semantic relationships among the latent vectors in the LDM. By replacing some of the computationally-intensive LDM inference iterations with our lightweight LatentCRF, we achieve a superior balance between quality, speed and diversity. We increase inference efficiency by 33% with no loss in image quality or diversity compared to the full LDM. LatentCRF is an easy add-on, which does not require modifying the LDM.
format Preprint
id arxiv_https___arxiv_org_abs_2412_18596
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LatentCRF: Continuous CRF for Efficient Latent Diffusion
Ranasinghe, Kanchana
Jayasumana, Sadeep
Veit, Andreas
Chakrabarti, Ayan
Glasner, Daniel
Ryoo, Michael S
Ramalingam, Srikumar
Kumar, Sanjiv
Computer Vision and Pattern Recognition
Latent Diffusion Models (LDMs) produce high-quality, photo-realistic images, however, the latency incurred by multiple costly inference iterations can restrict their applicability. We introduce LatentCRF, a continuous Conditional Random Field (CRF) model, implemented as a neural network layer, that models the spatial and semantic relationships among the latent vectors in the LDM. By replacing some of the computationally-intensive LDM inference iterations with our lightweight LatentCRF, we achieve a superior balance between quality, speed and diversity. We increase inference efficiency by 33% with no loss in image quality or diversity compared to the full LDM. LatentCRF is an easy add-on, which does not require modifying the LDM.
title LatentCRF: Continuous CRF for Efficient Latent Diffusion
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.18596