MOST-Net: A Memory Oriented Style Transfer Network for Face Sketch Synthesis

  title={MOST-Net: A Memory Oriented Style Transfer Network for Face Sketch Synthesis},
  author={Fan Ji and Muyi Sun and Xingqun Qi and Qi Li and Zhenan Sun},
  journal={2022 26th International Conference on Pattern Recognition (ICPR)},
  • Fan JiMuyi Sun Zhenan Sun
  • Published 8 February 2022
  • Computer Science
  • 2022 26th International Conference on Pattern Recognition (ICPR)
Face sketch synthesis has been widely used in multimedia entertainment and law enforcement. Despite the recent developments in deep neural networks, accurate and realistic face sketch synthesis is still a challenging task due to the diversity and complexity of human faces. Current image-to-image translation-based face sketch synthesis frequently encounters over-fitting problems when it comes to small-scale datasets. To tackle this problem, we present an end-to-end Memory Oriented Style Transfer… 

Figures and Tables from this paper



A Sketch-Transformer Network for Face Photo-Sketch Synthesis

A Sketch-Transformer network for face photo-sketch synthesis is proposed, which consists of three closely-related modules, including a multi-scale feature and position encoder for patch-level feature and positions embedding, a self-attention module for capturing long-range spatial dependency, and amulti-scale spatially-adaptive de-normalization decoder for image reconstruction.

Face Sketch Synthesis via Semantic-Driven Generative Adversarial Network

A novel Semantic-Driven Generative Adversarial Network (SDGAN) which embeds global structure-level style injection and local class-level knowledge re-weighting and dedicates to balance the contributions of different semantic classes is proposed.

Face Photo-Sketch Synthesis via Knowledge Transfer

Quantitative and qualitative evaluations illustrate that the proposed novel knowledge transfer framework can train a model which has superior performance using a small set of photosketch pairs and outperforms other state-of-the-art methods in generating face sketches with high visual quality and recognition ability.

End-to-End Photo-Sketch Generation via Fully Convolutional Representation Learning

This paper proposes a novel approach for photo-sketch generation, aiming to automatically transform face photos into detail-preserving personal sketches, and develops a fully convolutional network to learn the end-to-end photo-Sketch mapping.

Knowledge Distillation for Face Photo–Sketch Synthesis

This work investigates the effect of knowledge distillation (KD) on training neural networks for the face photo–sketch synthesis task and proposes an effective KD model to improve the performance of synthetic images and proposes a KD+ model that combines GANs with KD.

Toward Realistic Face Photo–Sketch Synthesis via Composition-Aided GANs

A novel composition-aided generative adversarial network (CA-GAN) for face photo-sketch synthesis is proposed, capable of generating both visually comfortable and identity-preserving face sketches/photos over a wide range of challenging data.

Deep Latent Low-Rank Representation for Face Sketch Synthesis

A novel face sketch synthesis framework based on deep latent low-rank representation (DLLRR) is proposed that can successfully provide the sufficient training data with the identity-specific information and synthesizes more clean and vivid face sketches.

Face Sketch Synthesis via Sparse Representation

A new face sketch synthesis method is presented, inspired by recent advances in sparse signal representation and neuroscience that human brain probably perceives images using high-level features which are sparse.

Markov Weight Fields for face sketch synthesis

A novel Markov Weight Fields (MWF) model is proposed that is capable of synthesizing new sketch patches and formulated into a convex quadratic programming (QP) problem to which the optimal solution is guaranteed.

Coupled information-theoretic encoding for face photo-sketch recognition

A new face descriptor based on coupled information-theoretic encoding is used to capture discriminative local face structures and to effectively match photos and sketches by reducing the modality gap at the feature extraction stage.