原创 Guided Image-to-Image Translation with Bi-Directional Feature Transformation(ICCV19)

不同於一般的image-to-image translation,本文主要針對帶guided信息的image-to-image translation,如Fig.1所示 第1行,給定一幅試衣圖像,指定guided信息爲pose關

原创 GANimation: Anatomically-aware Facial Animation from a Single Image(ECCV18)

3 Problem Formulation 定義輸入圖像Iyr∈RH×W×3\mathbf{I}_{\mathbf{y}_r}\in\mathbb{R}^{H\times W\times3}Iyr​​∈RH×W×3,yr=(y1,

原创 OpenCV Memo

獲取像素(i, j)的所有通道的值 unsigned char val = mat_variable.at<uchar>(i, j); // CV_8UC1 Vec3b &c = mat_variable.at<Vec3f>(i,

原创 TensorFlow Memo

multi-label使用的損失函數 loss = tf.losses.sigmoid_cross_entropy(tensor_label, tensor_logit)

原创 StarGAN v2: Diverse Image Synthesis for Multiple Domains

1. Introduction 定義domain和style Here, domain implies a set of images that can be grouped as a visually distinctive

原创 Paper Reading

【ICCV19】 Face De-occlusion using 3D Morphable Model and Generative Adversarial Network 使用3DMM生成一幅像是換臉一樣的圖像,然後與原圖拼接,

原创 One-shot Face Reenactment(BMVC19)

3 Approach 給定source face xsx_sxs​,包含了pose guidance,以及target face xtx_txt​,包含了reference appearance,學習的目標是生成一幅圖像包含xsx

原创 SinGAN: Learning a Generative Model from a Single Natural Image(ICCV19)

2. Method 學習的目標是an unconditional generative model that captures the internal statistics of a single training image

原创 Make a Face: Towards Arbitrary High Fidelity Face Manipulation(ICCV19)

3. Method 定義face image x∈Xx\in Xx∈X,給定target facial structural information ccc,學習一個mapping G\mathcal{G}G,將xxx轉換爲out

原创 AttGAN: Facial Attribute Editing by Only Changing What You Want(TIP19)

III. ATTRIBUTE GAN (ATTGAN) 前提:所有attribute都是binary型的 A. Testing Formulation 定義輸入圖像爲xa\mathbf{x^a}xa,包含nnn個attribut

原创 Age Progression and Regression with Spatial Attention Modules(AAAI20)

Method Problem Formulation 定義young face image爲Iy\mathbf{I}_yIy​,對應的age爲αy\bm{\alpha}_yαy​,是一個one-hot向量 給定目標age αo\b

原创 BeautyGAN: Instance-level Facial Makeup Transfer with Deep Generative Adversarial Network(ACMMM18)

3 OUR APPROACH: BEAUTYGAN non-makeup image domain A⊂RH×W×3A\subset \mathbb{R}^{H\times W\times 3}A⊂RH×W×3,makeup im

原创 PSGAN: Pose and Expression Robust Spatial-Aware GAN for Customizable Makeup Transfer(CVPR20)

3. PSGAN 3.1. Formulation source image domain XXX, reference image domain YYY domain XXX上有NNN個樣本,{xn}n=1,⋯ ,N,xn∈X\

原创 Learning Continuous Face Age Progression: A Pyramid of GANs(CVPR18擴展)

1 INTRODUCTION 本文是CVPR18的擴展 3 METHOD 3.1 Overview loss包括the traditional squared Euclidean loss、the GAN loss、the id

原创 MarioNETte: Few-shot Face Reenactment Preserving Identity of Unseen Targets(AAAI20)

MarioNETte Architecture Fig.2展示了MarioNETte的框架圖 給定driver image x\mathbf{x}x,一組target images {yi}i=1⋯K\left \{ \math