Learning Diverse Image Colorization

被引:98
作者
Deshpande, Aditya [1 ]
Lu, Jiajun [1 ]
Yeh, Mao-Chuang [1 ]
Chong, Min Jin [1 ]
Forsyth, David [1 ]
机构
[1] Univ Illinois, Champaign, IL 61820 USA
来源
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017) | 2017年
基金
美国国家科学基金会;
关键词
D O I
10.1109/CVPR.2017.307
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Colorization is an ambiguous problem, with multiple viable colorizations for a single grey-level image. However, previous methods only produce the single most probable colorization. Our goal is to model the diversity intrinsic to the problem of colorization and produce multiple colorizations that display long-scale spatial co-ordination. We learn a low dimensional embedding of color fields using a variational autoencoder (VAE). We construct loss terms for the VAE decoder that avoid blurry outputs and take into account the uneven distribution of pixel colors. Finally, we build a conditional model for the multi-modal distribution between grey-level image and the color field embeddings. Samples from this conditional model result in diverse colorization. We demonstrate that our method obtains better diverse colorizations than a standard conditional variational autoencoder (CVAE) model, as well as a recently proposed conditional generative adversarial network (cGAN).
引用
收藏
页码:2877 / 2885
页数:9
相关论文
共 31 条
  • [1] [Anonymous], 2016, Advances in Face Detection and Facial Image Analysis, DOI 10.1007/978-3-319-25958-1
  • [2] Batra D, 2012, LECT NOTES COMPUT SC, V7576, P1, DOI 10.1007/978-3-642-33715-4_1
  • [3] Bishop, 1994, MIXTURE DENSITY NETW, DOI DOI 10.1007/978-3-322-81570-58
  • [4] CHARPIAT G, 2008, P 10 EUR C COMP, V5304, P126
  • [5] Deep Colorization
    Cheng, Zezhou
    Yang, Qingxiong
    Sheng, Bin
    [J]. 2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, : 415 - 423
  • [6] Deshpande A., 2010, ICCV, P567
  • [7] Gregor K, 2015, PR MACH LEARN RES, V37, P1462
  • [8] Iizuka Satoshi, 2016, ACM T GRAPHICS P SIG, V35
  • [9] Ioffe Sergey, 2015, PROC INT C MACH LEAR, V37, P448, DOI DOI 10.48550/ARXIV.1502.03167
  • [10] Isola P., 2017, P IEEE C COMPUTER VI, P1125, DOI [DOI 10.1109/CVPR.2017.632, 10.1109/CVPR.2017.632]