Conditional Image Synthesis with Auxiliary Classifier GANs
By Augustus Odena, Chris Olah, Christopher Olah, Jonathon Shlens. 2016
3,545 citations recorded by the source corpus.
In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in 128 x 128 resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, 128 x 128 samples are more than twice as discriminable as artificially resized 32 x 32 samples. In addition, 84.7% of the classes have samples exhibiting diversity comparable to real ImageNet data.
Subfields: Biophysics, Computer Vision and Pattern Recognition, Media Technology, Generative Models
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Biophysics
- Jun-Yan Zhu
- Richard Zhang
- Yang Song
- Tianfu Wu
- Vladlen Koltun
- Stefano Ermon
- Alexei A. Efros
- Ming-Hsuan Yang
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Trevor Darrell
- Xiaogang Wang
- Jitendra Malik
- Yi Yang
- Bernt Schiele
In Media Technology
- Luc Van Gool
- Ming-Hsuan Yang
- Radu Timofte
- Chen Change Loy
- Chao Dong
- Wangmeng Zuo
- Dacheng Tao
- Lei Zhang
In Generative Models
- Jun-Yan Zhu
- Stefano Ermon
- Richard Zhang
- Eli Shechtman
- Yoshua Bengio
- Jiaming Song
- Alexei A. Efros
- Ming-Hsuan Yang
Citations in the corpus
- Are GANs Created Equal? A Large-Scale Study (2017, background): “Various flavors of GANs have been recently proposed, both purely unsupervised [9, 1, 10, 5] as well as conditional [20, 21].”
- Large Scale GAN Training for High Fidelity Natural Image Synthesis (2018, methodology): “In (Odena et al., 2017) it is provided to G by concatenating a 1-hot class vector to the noise vector, and the objective is modified to encourage conditional samples to maximize the corresponding class probability predicted by an auxiliary classifier.”
- Progressive Growing of GANs for Improved Quality, Stability, and Variation (2017, methodology): “We noticed that existing methods such as MS-SSIM (Odena et al., 2017) find large-scale mode collapses reliably but fail to react to smaller effects such as loss of variation in colors or textures, and they also do not directly assess image quality in terms of similarity to the training set.”
- Generative Adversarial Networks (2021, neutral): “One such technique is minibatch discrimination [14], which involves adding additional features to the Discriminator Network that allow it to compare multiple samples at once and differentiate between them.”
- Object-Centric Image Generation from Layouts (2020, methodology): “Generator and Discriminator losses We train the generator and patch discrim-inators using the adversarial hinge loss [30]: The object discriminator follows the AC-GAN framework [40], leading to L ACG and L ACD obj .”
- Improved Precision and Recall Metric for Assessing Generative Models (2019, background): “While metrics have been proposed to estimate the degree of variation, these have not seen widespread use as they are subjective [1], domain specific [19], or not reliable enough [22].”
- Semantic Image Synthesis With Spatially-Adaptive Normalization (2019, background): “For example, class-conditional models [3, 36, 37, 39, 41] learn to synthesize images given category labels.”
- Improved Training of Wasserstein GANs (2017, methodology): “When we add label information (using the procedure in [19]), the same architecture outperforms all other published models except for SGAN.”
- Unrolled Generative Adversarial Networks (2016, methodology): “…functions, without explicitly computing likelihoods or performing inference, and while targeting more mode-seeking divergences as made GANs extremely successful in image generation (Odena et al., 2016; Salimans et al., 2016; Radford et al., 2015), and image super resolution (Ledig et al., 2016).”
- DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-To-Image Synthesis (2019, background): “With the recent successes of Variational Autoencoders (VAEs) [9] and GANs [4], a large number of methods have been proposed to handle generation [14, 17, 28, 1] and domain adaptation task [25, 32].”
- Self-Supervised GANs via Auxiliary Rotation Loss (2018, methodology): “A key technique in these settings is conditioning [9, 20, 21, 22] whereby both the generator and discriminator have access to labeled data.”
- Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step (2017, methodology): “Since our models are completely unsupervised, we do not compute the similarity across samples of the same class as in [18].”
- Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, methodology): “AC-GANs add the ability to generate labeled samples.”
- The Numerics of GANs (2017, background): “Other approaches that try to stabilize training, try to design an easy-to-optimize architecture [23, 21] or make use of additional labels [23, 17].”
- Generalization and Equilibrium in Generative Adversarial Nets (GANs) (2017, methodology): “…method on image generation tasks: MNIST dataset [LeCun et al., 1998] of hand-written digits and the CelebA [Liu et al., 2015] dataset Liu, 2016] 6.35 Improved GAN [Salimans et al., 2016] 8.09 ± 0.07 AC-GAN [Odena et al., 2016] 8.25 ± 0.07 S-GAN (best variant in [Huang et al., 2017 of human faces.”
- Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis (2017, methodology): “We build an auxiliary task into our system following the ACGAN [2] formulation.”