mlsanityData to 7 Oct 2026

Toward Spatially Unbiased Generative Models

arXiv 2108.01285 ↗First posted 3 Aug 2021cs.LG cs.CVcited by 17

Not on the map. Why?

Abstract (arXiv)

Recent image generation models show remarkable generation performance. However, they mirror strong location preference in datasets, which we call spatial bias. Therefore, generators render poor samples at unseen locations and scales. We argue that the generators rely on their implicit positional encoding to render spatial content. From our observations, the generator's implicit positional encoding is translation-variant, making the generator spatially biased. To address this issue, we propose injecting explicit positional encoding at each scale of the generator. By learning the spatially unbiased generator, we facilitate the robust use of generators in multiple tasks, such as GAN inversion, multi-scale generation, generation of arbitrary sizes and aspect ratios. Furthermore, we show that our method can also be applied to denoising diffusion probabilistic models.

Key facts. Cited by 17 papers in the map (data through 7 Oct 2026). Builds on 41 of its 46 references that are in the map. First posted on arXiv 3 Aug 2021.

Builds on

41 of 46 references.

References · 41

  1. 2017Attention Is All You Needcited by 59,932
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  4. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  5. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  6. 2020End-to-End Object Detection with Transformerscited by 7,384
  7. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  8. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  9. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,736
  10. 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
  11. 2017Progressive Growing of GANs for Improved Quality, Stability, and Variationcited by 2,967
  12. 2018Large Scale GAN Training for High Fidelity Natural Image Synthesiscited by 2,609
29 more
  1. 2018Group Normalizationcited by 1,514
  2. 2015LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loopcited by 1,350
  3. 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,339
  4. 2021Alias-Free Generative Adversarial Networkscited by 1,181
  5. 2019Interpreting the Latent Space of GANs for Semantic Face Editingcited by 691
  6. 2019Image2StyleGAN: How to Embed Images Into the StyleGAN Latent Space?cited by 675
  7. 2018Which Training Methods for GANs do actually Converge?cited by 672
  8. 2020WaveGrad: Estimating Gradients for Waveform Generationcited by 620
  9. 2020Closed-Form Factorization of Latent Semantics in GANscited by 407
  10. 2018Assessing Generative Models via Precision and Recallcited by 401
  11. 2019Making Convolutional Networks Shift-Invariant Againcited by 390
  12. 2020PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Modelscited by 358
  13. 2016Generative Visual Manipulation on the Natural Image Manifoldcited by 348
  14. 2019SinGAN: Learning a Generative Model from a Single Natural Imagecited by 348
  15. 2018An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solutioncited by 302
  16. 2019Learning De-biased Representations with Biased Representationscited by 199
  17. 2018Why do deep convolutional networks generalize so poorly to small image transformations?cited by 198
  18. 2019Image Processing Using Multi-Code GAN Priorcited by 185
  19. 2020How Much Position Information Do Convolutional Neural Networks Encode?cited by 171
  20. 2016Semantic Image Inpainting with Deep Generative Modelscited by 156
  21. 2020On Translation Invariance in CNNs: Convolutional Layers can Exploit Absolute Spatial Locationcited by 141
  22. 2016Neural Photo Editing with Introspective Adversarial Networkscited by 109
  23. 2018Bias and Generalization in Deep Generative Models: An Empirical Studycited by 83
  24. 2020Transforming and Projecting Images into Class-conditional Generative Networkscited by 76
  25. 2020Positional Encoding as Spatial Inductive Bias in GANscited by 69
  26. 2019COCO-GAN: Generation by Parts via Conditional Coordinatingcited by 67
  27. 2020Mind the Pad -- CNNs can Develop Blind Spotscited by 41
  28. 2020Shift Equivariance in Object Detectioncited by 13
  29. 2020A Note on Data Biases in Generative Modelscited by 13

Built on by

Cited by year

Citing papers per year: 2021: 3; 2022: 5; 2023: 6; 2024: 3.

17 papers.

Sort:

By area:

  1. 2021ILVR: Conditioning Method for Denoising Diffusion Probabilistic Modelscited by 634
  2. 2021StyleSwin: Transformer-based GAN for High-resolution Image Generationcited by 152
  3. 2022Any-resolution Training for High-resolution Image Synthesiscited by 72
  4. 2023LiDAR Data Synthesis with Denoising Diffusion Probabilistic Modelscited by 53
  5. 2022Learning to Generate 3D Shapes from a Single Examplecited by 31
  6. 2024Arbitrary-Scale Image Generation and Upsampling using Latent Diffusion Model and Implicit Neural Decodercited by 25
  7. 2021LTT-GAN: Looking Through Turbulence by Inverting GANscited by 21
  8. 2022Arbitrary-Scale Image Synthesiscited by 20
  9. 2024The Crystal Ball Hypothesis in diffusion models: Anticipating object positions from initial noisecited by 18
  10. 2023OPE-SR: Orthogonal Position Encoding for Designing a Parameter-free Upsampling Module in Arbitrary-scale Image Super-Resolutioncited by 17
  11. 2023UnitedHuman: Harnessing Multi-Source Data for High-Resolution Human Generationcited by 9
  12. 2023Zone Evaluation: Revealing Spatial Bias in Object Detectioncited by 7
5 more
  1. 2022Towards Universal Texture Synthesis by Combining Texton Broadcasting with Noise Injection in StyleGAN-2cited by 3
  2. 2022Unveiling The Mask of Position-Information Pattern Through the Mist of Image Featurescited by 3
  3. 2023DARTS: Double Attention Reference-based Transformer for Super-resolutioncited by 2
  4. 2024HiTSR: A Hierarchical Transformer for Reference-based Super-Resolutioncited by 1
  5. 2023Improving Translation Invariance in Convolutional Neural Networks with Peripheral Prediction Paddingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Position Forcing: Self-Conditioning 3D Generation
  2. 2025Towards Spatially Consistent Image Generation: On Incorporating Intrinsic Scene Properties into Diffusion Models
  3. 2026AsyncPatch Diffusion: spatially-flexible image generation
  4. 2026PE-Field 4D: Video Generation Models as Canvas
  5. 2026StructDiff: A Structure-Preserving and Spatially Controllable Diffusion Model for Single-Image Generation
  6. 2025Autoregressive Image Generation with Linear Complexity: A Spatial-Aware Decay Perspective
  7. 2023Diffusion Models Generate Images Like Painters: an Analytical Theory of Outline First, Details Later
  8. 2024Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation