Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
149 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

MetaFBP: Learning to Learn High-Order Predictor for Personalized Facial Beauty Prediction (2311.13929v1)

Published 23 Nov 2023 in cs.CV

Abstract: Predicting individual aesthetic preferences holds significant practical applications and academic implications for human society. However, existing studies mainly focus on learning and predicting the commonality of facial attractiveness, with little attention given to Personalized Facial Beauty Prediction (PFBP). PFBP aims to develop a machine that can adapt to individual aesthetic preferences with only a few images rated by each user. In this paper, we formulate this task from a meta-learning perspective that each user corresponds to a meta-task. To address such PFBP task, we draw inspiration from the human aesthetic mechanism that visual aesthetics in society follows a Gaussian distribution, which motivates us to disentangle user preferences into a commonality and an individuality part. To this end, we propose a novel MetaFBP framework, in which we devise a universal feature extractor to capture the aesthetic commonality and then optimize to adapt the aesthetic individuality by shifting the decision boundary of the predictor via a meta-learning mechanism. Unlike conventional meta-learning methods that may struggle with slow adaptation or overfitting to tiny support sets, we propose a novel approach that optimizes a high-order predictor for fast adaptation. In order to validate the performance of the proposed method, we build several PFBP benchmarks by using existing facial beauty prediction datasets rated by numerous users. Extensive experiments on these benchmarks demonstrate the effectiveness of the proposed MetaFBP method.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (53)
  1. The automatic measurement of facial beauty. In IEEE SMC, Vol. 4. 2644–2647.
  2. Convolutional neural networks for speech recognition. IEEE/ACM Trans. Audio Speech Lang. 22, 10 (2014), 1533–1545.
  3. Andrea Bottino and Aldo Laurentini. 2012. The intrinsic dimensionality of attractiveness: A study in face profiles. In CIARP. 59–66.
  4. Memory matching networks for one-shot image recognition. In CVPR. 4080–4088.
  5. Label Matching Semi-Supervised Object Detection. In CVPR. 14381–14390.
  6. Data-driven facial beauty analysis: prediction, retrieval and manipulation. IEEE Trans. Affect. Comput. 9, 2 (2016), 205–216.
  7. Customized Automatic Face Beautification. In ICASSP. 1–5.
  8. Unsupervised image classification for deep representation learning. In ECCV. 430–446.
  9. All you need is a few shifts: Designing efficient convolutional neural networks for image classification. In CVPR. 7241–7250.
  10. Distribution-oriented aesthetics assessment with semantic-aware hybrid network. TMM 21, 5 (2018), 1209–1220.
  11. Imagenet: A large-scale hierarchical image database. In CVPR. 248–255.
  12. Personalized image aesthetics assessment. In ACM CIKM. 2043–2046.
  13. Label distribution-based facial attractiveness computation by deep residual learning. TMM 20, 8 (2017), 2196–2208.
  14. Model-agnostic meta-learning for fast adaptation of deep networks. In ICML. 1126–1135.
  15. Deep residual learning for image recognition. In CVPR. 770–778.
  16. A humanlike predictor of facial attractiveness. In NeurIPS. 649–656.
  17. Konstantinos Kalais and Sotirios Chatzis. 2022. Stochastic Deep Networks with Linear Competing Units for Model-Agnostic Meta-Learning. In ICML. 10586–10597.
  18. Segment anything. arXiv preprint arXiv:2304.02643 (2023).
  19. ImageNet Classification with Deep Convolutional Neural Networks. In NeurIPS, Vol. 25.
  20. Finding task-relevant features for few-shot learning by category traversal. In CVPR. 1–10.
  21. Personality driven multi-task learning for image aesthetic assessment. In ICME. 430–435.
  22. Personality-assisted multi-task learning for generic and personalized image aesthetics assessment. TIP 29 (2020), 3898–3910.
  23. Transductive aesthetic preference propagation for personalized image aesthetics assessment. In ACM MM. 896–904.
  24. SCUT-FBP5500: A diverse benchmark dataset for multi-paradigm facial beauty prediction. In ICPR. 1598–1603.
  25. Region-aware scattering convolution networks for facial beauty prediction. In ICIP. 2861–2865.
  26. R2-resnext: A resnext-based regression model with relative ranking for facial beauty prediction. In ICPR. 85–90.
  27. Regression Guided by Relative Ranking Using Convolutional Neural Network (R3CNN) for Facial Beauty Prediction. IEEE Trans. on Affec. Comput. 13, 1 (2019), 122–134.
  28. Attribute-Aware Convolutional Neural Networks for Facial Beauty Prediction. In IJCAI. 847–853.
  29. Microsoft coco: Common objects in context. In ECCV. 740–755.
  30. Tsendsuren Munkhdalai and Hong Yu. 2017. Meta networks. In ICML. 2554–2563.
  31. Latent embedding feedback and discriminative features for zero-shot classification. In ECCV. 479–495.
  32. On first-order meta-learning algorithms. arXiv preprint arXiv:1803.02999 (2018).
  33. Learning transferable visual models from natural language supervision. In ICML. 8748–8763.
  34. Faster R-CNN: towards real-time object detection with region proposal networks. TPAMI 39, 6 (2016), 1137–1149.
  35. Some like it hot-visual guidance for preference prediction. In CVPR. 5553–5561.
  36. Meta-learning with memory-augmented neural networks. In ICML. 1842–1850.
  37. Prototypical networks for few-shot learning. NeurIPS 30 (2017).
  38. Meta-transfer learning for few-shot learning. In CVPR. 403–412.
  39. Learning to compare: Relation network for few-shot learning. In CVPR. 1199–1208.
  40. Sebastian Thrun and Lorien Pratt. 1998. Learning to learn: Introduction and overview. In Learning to learn. Springer, 3–17.
  41. Matching networks for one shot learning. NeurIPS 29 (2016), 3630–3638.
  42. [Demo paper] exploring attractive faces: General versus personal preferences. In ICMEW. 1–2.
  43. Zero-shot learning via class-conditioned deep generative models. In AAAI, Vol. 32.
  44. Meta-learning perspective for personalized image aesthetics assessment. In ICIP. 1875–1879.
  45. Generalizing from a few examples: A survey on few-shot learning. ACM Comput. Surv. 53, 3 (2020), 1–34.
  46. Jacob Whitehill and Javier R Movellan. 2008. Personalized facial attractiveness prediction. In FG. IEEE, 1–7.
  47. Scut-fbp: A benchmark dataset for facial beauty perception. In SMC. 1821–1826.
  48. Hansi Yang and James Kwok. 2022. Efficient Variance Reduction for Meta-Learning. In ICML. 25070–25095.
  49. Personalized image aesthetics assessment with rich attributes. In CVPR. 19861–19869.
  50. Tapnet: Neural network augmented with task-adaptive projection for few-shot learning. In ICML. 7115–7123.
  51. A gated peripheral-foveal convolutional neural network for unified image aesthetic prediction. TMM 21, 11 (2019), 2815–2826.
  52. Personalized image aesthetics assessment via meta-learning with bilevel gradient optimization. IEEE Trans. on Cyber. 52, 3 (2020), 1798–1811.
  53. Learning Personalized Image Aesthetics From Subjective and Objective Attributes. TMM 25 (2023), 179–190.
Citations (3)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now