Analyzing the Feature Extractor Networks for Face Image Synthesis
Not on the map. Why?
Abstract (arXiv)
Advancements like Generative Adversarial Networks have attracted the attention of researchers toward face image synthesis to generate ever more realistic images. Thereby, the need for the evaluation criteria to assess the realism of the generated images has become apparent. While FID utilized with InceptionV3 is one of the primary choices for benchmarking, concerns about InceptionV3's limitations for face images have emerged. This study investigates the behavior of diverse feature extractors -- InceptionV3, CLIP, DINOv2, and ArcFace -- considering a variety of metrics -- FID, KID, Precision&Recall. While the FFHQ dataset is used as the target domain, as the source domains, the CelebA-HQ dataset and the synthetic datasets generated using StyleGAN2 and Projected FastGAN are used. Experiments include deep-down analysis of the features: L2 normalization, model attention during extraction, and domain distributions in the feature space. We aim to give valuable insights into the behavior of feature extractors for evaluating face image synthesis methodologies. The code is publicly available at https://github.com/ThEnded32/AnalyzingFeatureExtractors.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 29 of its 35 references that are in the map. First posted on arXiv 4 Jun 2024.
Builds on
29 of 35 references.
References · 29
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
- 2020Denoising Diffusion Probabilistic Modelscited by 20,851
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
- 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
- 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
- 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,510
- 2014Going Deeper with Convolutionscited by 4,231
- 2013Auto-Encoding Variational Bayescited by 3,980
- 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
17 more
- 2017Progressive Growing of GANs for Improved Quality, Stability, and Variationcited by 2,967
- 2016Improved Techniques for Training GANscited by 2,938
- 2014Conditional Generative Adversarial Netscited by 2,152
- 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
- 2023Vision Transformers Need Registerscited by 844
- 2016MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognitioncited by 621
- 2022The Role of ImageNet Classes in Fr'echet Inception Distancecited by 211
- 2021Projected GANs Converge Fastercited by 204
- 2020Understanding How Dimension Reduction Tools Work: An Empirical Approach to Deciphering t-SNE, UMAP, TriMAP, and PaCMAP for Data Visualizationcited by 170
- 2023IDiff-Face: Synthetic-based Face Recognition through Fizzy Identity-Conditioned Diffusion Modelscited by 72
- 2019HYPE: A Benchmark for Human eYe Perceptual Evaluation of Generative Modelscited by 59
- 2022A Study on the Evaluation of Generative Modelscited by 35
- 2022Learning the Degradation Distribution for Blind Image Super-Resolutioncited by 27
- 2023Bellman Optimal Stepsize Straightening of Flow-Matching Modelscited by 11
- 2018Use of Neural Signals to Evaluate the Quality of Generative Adversarial Network Performance in Facial Image Generationcited by 7
- 2023Revisiting the Evaluation of Image Synthesis with GANscited by 7
- 2023Toward responsible face datasets: modeling the distribution of a disentangled latent space for sampling face images from demographic groupscited by 6
Built on by
Citing papers per year: 2024: 1.
1 paper.