Perceptual Loss

الخسارة الإدراكية

دالة خسارة تقيس الفرق بين صورتين في فضاء السمات العميقة لشبكة مُدرَّبة مسبقاً (مثل VGG)، بدلاً من مقارنة البكسلات مباشرة، فتُنتج إعادة بناء تبدو صحيحة للعين البشرية.

A loss function measuring the difference between two images in the deep feature space of a pretrained network (e.g. VGG), rather than comparing pixels directly, producing reconstructions that look correct to the human eye.

Also translated asخسارة السمات العميقة، دالة الخسارة الإدراكية

First appears in this corpus in: Very Deep Convolutional Networks for Large-Scale Image Recognition (2014)

Appears in these papers

High-Resolution Image Synthesis with Latent Diffusion Models2022 in the sky ✦
Image-to-Image Translation with Conditional Adversarial Networks2017 in the sky ✦
Image Style Transfer Using Convolutional Neural Networks2016 in the sky ✦
Image Style Transfer Using Convolutional Neural Networks2016 in the sky ✦
A Style-Based Generator Architecture for Generative Adversarial Networks2019 in the sky ✦
Analyzing and Improving the Image Quality of StyleGAN2020 in the sky ✦
Analyzing and Improving the Image Quality of StyleGAN2020 in the sky ✦
Very Deep Convolutional Networks for Large-Scale Image Recognition2014 in the sky ✦