image-generation

Image Harmonization

3260 papers • 126 benchmarks • 313 datasets

Image harmonization aims to modify the color of the composited region with respect to the specific background.

(Image credit: Papersgraph)

Benchmarks

These leaderboards are used to track progress in image-generation

Trend

Dataset

Best Model

Actions

iHarmony4

$HAdobe5k(1024$\times$1024)$

HAdobe5k(1024$\times$1024)

FFHQH

Libraries

i

Use these libraries to find image-generation models and implementations

Datasets

iHarmony4

FFHQH

Subtasks

No subtasks available.

Most implemented papers

PHNet: Patch-based Normalization for Portrait Harmonization

A. Kapitanov, Elizaveta Petrova, Karen Efremyan, Evgeny Kaskov•Mon Feb 26 2024

This work presents a patch-based harmonization network consisting of novel Patch-based normalization (PN) blocks and a feature extractor based on statistical color transfer and achieves state-of-the-art results on the iHarmony4 dataset.

1

Content

FFHQH

0

Paper Graph

DoveNet: Deep Image Harmonization via Domain Verification

Wenyan Cong, Jianfu Zhang, Li Niu, Liu Liu, Zhixin Ling, Weiyuan Li, Liqing Zhang•Tue Nov 26 2019

This work contributes an image harmonization dataset iHarmony4 by generating synthesized composite images based on COCO (resp., Adobe5k, Flickr, day2night) dataset, leading to the HCOCO sub-dataset, and proposes a new deep image harmonized method DoveNet using a novel domain verification discriminator.

246 0

Paper Graph

Image Harmonization Dataset iHarmony4: HCOCO, HAdobe5k, HFlickr, and Hday2night.

Wenyan Cong, Li Niu, Liu Liu, Zhixin Ling, Weiyuan Li, Liqing Zhang, Jianfu Zhang•Tue Aug 27 2019

This work contributes an image harmonization dataset iHarmony4 by generating synthesized composite images based on existing COCO (resp., Adobe5k, day2night) dataset, leading to the HCOCO sub-dataset.

6 0

Paper Graph

Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization

Chi-Man Pun, Xiaodong Cun, Jing Liang•Sun Sep 12 2021

A novel spatial-separated curve rendering network for efficient and high-resolution image harmonization for the first time and reduces more than 90% parameters compared with previous methods but still achieves the state-of-the-art performance on both synthesized iHarmony4 and real-world DIH test sets.

48 0

Paper Graph

Foreground-aware Semantic Representations for Image Harmonization

Konstantin Sofiiuk, Anton Konushin, Polina Popenova•Sun May 31 2020

This work proposes a novel architecture to utilize the space of high-level features learned by a pre-trained classification network to harmonize composites and sets up a new state-of-the-art in terms of MSE and PSNR metrics.

89 0

Paper Graph

Making Images Real Again: A Comprehensive Survey on Deep Image Composition

Wenyan Cong, Li Niu, Liu Liu, Liqing Zhang, Bo Zhang, Jing Liang, Yan Hong•Sun Jun 27 2021

A comprehensive survey over the sub-tasks and combined task of image composition is conducted, which summarizes the existing methods, available datasets, and common evaluation metrics.

96 0

Paper Graph

Improving the Harmony of the Composite Image by Spatial-Separated Attention Module

Chi-Man Pun, Xiaodong Cun•Sun Jul 14 2019

This work proposes a novel attention module named Spatial-Separated Attention Module (S2AM) and designs a novel image harmonization framework by inserting the S2AM in the coarser low-level features of the Unet structure by two different ways.

154 0

Paper Graph

Generative Hierarchical Features from Synthesizing Images

Bolei Zhou, Ceyuan Yang, Yinghao Xu, Yujun Shen, Jiapeng Zhu•Sun Jul 19 2020

This work considers the pre-trained StyleGAN generator as a learned loss function and utilizes its layer-wise representation to train a novel hierarchical encoder, termed as Generative Hierarchical Feature (GH-Feat), which has strong transferability to both generative and discriminative tasks.

125 0

Paper Graph

Bargainnet: Background-Guided Domain Translation for Image Harmonization

Wenyan Cong, Li Niu, Liqing Zhang, Jianfu Zhang, Jing Liang•Fri Sep 18 2020

This work forms image harmonization task as background-guided domain translation, which is regulated by well-tailored triplet losses, and uses a domain code extractor to capture the background domain information to guide the foreground harmonization.

103 0

Paper Graph

Adding a benchmark result helps the community track progress.