FID
مسافة فريشيه للبداية
مقياس كمّي يُقيّم جودة الصور المُولَّدة بحساب المسافة بين توزيعات السمات للصور الحقيقية والمُولَّدة — كلما انخفضت القيمة تحسّنت الجودة.
A quantitative metric evaluating generated image quality by computing the distance between feature distributions of real and generated images — lower values indicate better quality.
Also translated asFID، مقياس FID
First appears in this corpus in: Progressive Growing of GANs for Improved Quality, Stability, and Variation (2018)
Appears in these papers
- Consistency Models2023in the sky ✦
- Denoising Diffusion Implicit Models2021in the sky ✦
- Denoising Diffusion Probabilistic Models2020in the sky ✦
- Diffusion Models Beat GANs on Image Synthesis2021in the sky ✦
- Flow Matching for Generative Modeling2023in the sky ✦
- Imagen: Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding2022in the sky ✦
- High-Resolution Image Synthesis with Latent Diffusion Models2022in the sky ✦
- Progressive Growing of GANs for Improved Quality, Stability, and Variation2018in the sky ✦
- A Style-Based Generator Architecture for Generative Adversarial Networks2019in the sky ✦
- Analyzing and Improving the Image Quality of StyleGAN2020in the sky ✦