Smoothing the Disentangled Latent Style Space for Unsupervised Image-to-Image Translation

A comprehensive survey on semantic facial attribute editing using generative adversarial networks

“…Attribute interpolation In semantic facial editing, a face image is edited by changing an attribute from an initial value to a target value. In some cases, intermediate states of an attribute can also result in valid face images [107]. Genrating such intermediate images is known as attribute interpolation and is done in a similar way as latent space interpolation explained earlier.…”

Contrastive Monotonic Pixel-Level Modulation

“…As one of the most popular and successful schemes, disentangling the content and style representation exhibits a great success on a wealth of attribute and style translation tasks [36,26,13,73,31], and enables a continuous translation by interpolating between two latent vectors [31,13]. Recently, works on continuous cross-domain translation further refine the quality of intermediate images by introducing an interpolation discriminator [71,43], constraining the intermediate results with discriminators from both sides [18,51], or by exploiting the path of interpolation and translation manifold [11,56,44].…”

Predict, Prevent, and Evaluate: Disentangled Text-Driven Image Manipulation Empowered by Pre-Trained Vision-Language Model

“…Due to the costly labor, a key challenge of such works is reducing the supervision for learning the desired disentanglement. Therefore, weakly-supervised and unsupervised methods have been explored [8,10,23,27]. Despite progress, all these methods are trained for a fixed set of attributes, thus supporting limited numbers of manipulations.…”