Face photo–sketch synthesis via intra-domain enhancement

DALS: Diffusion-Based Artistic Landscape Sketch

“…Various sketch synthesizing methods can be categorized into four groups: (1) pre-deep learning methods [1][2][3]28] that use mathematical principles to extract prominent lines and to create hatching patterns using kernel-based techniques, (2) general deep learning schemes [4][5][6][7][8][9][10][11] including neural style transfer (NST) or generative adversarial networks (GAN) that synthesize sketch images from input images, (3) specialized deep learning schemes [12][13][14][15][16][17][18][19][20] designed for sketch synthesis, such as portrait sketch synthesis, forensic facial sketch synthesis, and architectural sketch synthesis, and (4) diffusion model-based methods [24][25][26][29][30][31][32][33][34], which are relatively less explored for sketch synthesis due to the dearth of high-quality sketch datasets.…”

“…This technique introduces an innovative approach leveraging existing knowledge from similar domains. Peng et al [16] introduced a two-stage strategy for face sketch synthesis, while the Dual Conditional Normalization Pyramid (DCNP) [17] was proposed to synthesize face sketch images using reference samples.…”

“…Artistic sketch is a genre of fine arts that has long been beloved. Many researchers have proposed various schemes for synthesizing sketches from images [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. However, many of them focus on synthesizing portrait sketches [10,11,[13][14][15]20].…”

Intelligent face sketch recognition system using shearlet transform and convolutional neural network model

“…An intra-domain enhancement (IDE) method are represented in [21], which targeting both modality gap and quality issues within the same domain for face photo-sketch synthesis. Extensive experiments on public face sketch databases affirm the IDE-based approach's superiority over current state-of-the-art methods, particularly in detail preservation.…”