Jingtang Liang*, Xiaodong Cun*, Chi-Man Pun, Jue Wang
Paper: https://arxiv.org/abs/2109.05750
Demos: https://github.com/vinthony/S2CRNet-demos
Image harmonization aims to modify the color of the composited region with respect to the specific background. Previous works model this task as a pixel-wise image-to-image translation using UNet family structures. However, the model size and computational cost limit the performability of their models on edge devices and higher-resolution images. To this end, we propose a novel spatial-separated curve rendering network(S2CRNet) for efficient and high-resolution image harmonization for the first time. In S2CRNet, we firstly extract the spatial-separated embeddings from the thumbnails of the masked foreground and background individually. Then, we design a curve rendering module(CRM), which learns and combines the spatial-specific knowledge using linear layers to generate the parameters of the pixel-wise curve mapping in the foreground region. Finally, we directly render the original high-resolution images using the learned color curve. Besides, we also make two extensions of the proposed framework via the Cascaded-CRM and Semantic-CRM for cascaded refinement and semantic guidance, respectively. Experiments show that the proposed method 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 set. Moreover, our method can work smoothly on higher resolution images in real-time which is more than 10× faster than the existing methods.
@misc{liang2021spatialseparated,
title={Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization},
author={Jingtang Liang and Xiaodong Cun and Chi-Man Pun and Jue Wang},
year={2021},
eprint={2109.05750},
archivePrefix={arXiv},
primaryClass={cs.CV}
}