Corpus ID: 211259260

Reliable Fidelity and Diversity Metrics for Generative Models

@article{Naeem2020ReliableFA,
  title={Reliable Fidelity and Diversity Metrics for Generative Models},
  author={Muhammad Ferjad Naeem and Seong Joon Oh and Youngjung Uh and Yunjey Choi and Jaejun Yoo},
  journal={ArXiv},
  year={2020},
  volume={abs/2002.09797}
}
Devising indicative evaluation metrics for the image generation task remains an open problem. The most widely used metric for measuring the similarity between real and generated images has been the Frechet Inception Distance (FID) score. Because it does not differentiate the fidelity and diversity aspects of the generated images, recent papers have introduced variants of precision and recall metrics to diagnose those properties separately. In this paper, we show that even the latest version of… Expand
How Faithful is your Synthetic Data? Sample-level Metrics for Evaluating and Auditing Generative Models
Equivalence of several curves assessing the similarity between probability distributions
Few-shot Image Generation via Cross-domain Correspondence
Instance Selection for GANs
Adversarial Text-to-Image Synthesis: A Review
The Spatially-Correlative Loss for Various Image Translation Tasks
Few-shot Compositional Font Generation with Dual Memory
...
1
2
3
...

References

SHOWING 1-10 OF 26 REFERENCES
Improved Precision and Recall Metric for Assessing Generative Models
Assessing Generative Models via Precision and Recall
A note on the evaluation of generative models
The Unreasonable Effectiveness of Deep Features as a Perceptual Metric
Large Scale GAN Training for High Fidelity Natural Image Synthesis
Improved Techniques for Training GANs
A Style-Based Generator Architecture for Generative Adversarial Networks
Pros and Cons of GAN Evaluation Measures
  • A. Borji
  • Computer Science, Mathematics
  • Comput. Vis. Image Underst.
  • 2019
Deep Image Prior
...
1
2
3
...