Photo2cartoon - 人像卡通化探索项目 (photo-to-cartoon translation project)

Overview

人像卡通化 (Photo to Cartoon)

中文版 | English Version

该项目为小视科技卡通肖像探索项目。您可使用微信扫描下方二维码或搜索“AI卡通秀”小程序体验卡通化效果。

也可以前往我们的ai开放平台进行在线体验:https://ai.minivision.cn/#/coreability/cartoon

技术交流QQ群:937627932

Updates

简介

人像卡通风格渲染的目标是,在保持原图像ID信息和纹理细节的同时,将真实照片转换为卡通风格的非真实感图像。我们的思路是,从大量照片/卡通数据中习得照片到卡通画的映射。一般而言,基于成对数据的pix2pix方法能达到较好的图像转换效果,但本任务的输入输出轮廓并非一一对应,例如卡通风格的眼睛更大、下巴更瘦;且成对的数据绘制难度大、成本较高,因此我们采用unpaired image translation方法来实现。

Unpaired image translation流派最经典方法是CycleGAN,但原始CycleGAN的生成结果往往存在较为明显的伪影且不稳定。近期的论文U-GAT-IT提出了一种归一化方法——AdaLIN,能够自动调节Instance Norm和Layer Norm的比重,再结合attention机制能够实现精美的人像日漫风格转换。

与夸张的日漫风不同,我们的卡通风格更偏写实,要求既有卡通画的简洁Q萌,又有明确的身份信息。为此我们增加了Face ID Loss,使用预训练的人脸识别模型提取照片和卡通画的ID特征,通过余弦距离来约束生成的卡通画。

此外,我们提出了一种Soft-AdaLIN(Soft Adaptive Layer-Instance Normalization)归一化方法,在反规范化时将编码器的均值方差(照片特征)与解码器的均值方差(卡通特征)相融合。

模型结构方面,在U-GAT-IT的基础上,我们在编码器之前和解码器之后各增加了2个hourglass模块,渐进地提升模型特征抽象和重建能力。

由于实验数据较为匮乏,为了降低训练难度,我们将数据处理成固定的模式。首先检测图像中的人脸及关键点,根据人脸关键点旋转校正图像,并按统一标准裁剪,再将裁剪后的头像输入人像分割模型去除背景。

Start

安装依赖库

项目所需的主要依赖库如下:

  • python 3.6
  • pytorch 1.4
  • tensorflow-gpu 1.14
  • face-alignment
  • dlib
  • onnxruntime

Clone:

git clone https://github.com/minivision-ai/photo2cartoon.git
cd ./photo2cartoon

下载资源

谷歌网盘 | 百度网盘 提取码:y2ch

  1. 人像卡通化预训练模型:photo2cartoon_weights.pt(20200504更新),存放在models路径下。
  2. 头像分割模型:seg_model_384.pb,存放在utils路径下。
  3. 人脸识别预训练模型:model_mobilefacenet.pth,存放在models路径下。(From: InsightFace_Pytorch
  4. 卡通画开源数据:cartoon_data,包含trainBtestB
  5. 人像卡通化onnx模型:photo2cartoon_weights.onnx 谷歌网盘,存放在models路径下。

测试

将一张测试照片(亚洲年轻女性)转换为卡通风格:

python test.py --photo_path ./images/photo_test.jpg --save_path ./images/cartoon_result.png

测试onnx模型

python test_onnx.py --photo_path ./images/photo_test.jpg --save_path ./images/cartoon_result.png

训练

1.数据准备

训练数据包括真实照片和卡通画像,为降低训练复杂度,我们对两类数据进行了如下预处理:

  • 检测人脸及关键点。
  • 根据关键点旋转校正人脸。
  • 将关键点边界框按固定的比例扩张并裁剪出人脸区域。
  • 使用人像分割模型将背景置白。

我们开源了204张处理后的卡通画数据,您还需准备约1000张人像照片(为匹配卡通数据,尽量使用亚洲年轻女性照片,人脸大小最好超过200x200像素),使用以下命令进行预处理:

python data_process.py --data_path YourPhotoFolderPath --save_path YourSaveFolderPath

将处理后的数据按照以下层级存放,trainAtestA中存放照片头像数据,trainBtestB中存放卡通头像数据。

├── dataset
    └── photo2cartoon
        ├── trainA
            ├── xxx.jpg
            ├── yyy.png
            └── ...
        ├── trainB
            ├── zzz.jpg
            ├── www.png
            └── ...
        ├── testA
            ├── aaa.jpg 
            ├── bbb.png
            └── ...
        └── testB
            ├── ccc.jpg 
            ├── ddd.png
            └── ...

2.训练

重新训练:

python train.py --dataset photo2cartoon

加载预训练参数:

python train.py --dataset photo2cartoon --pretrained_weights models/photo2cartoon_weights.pt

多GPU训练(仍建议使用batch_size=1,单卡训练):

python train.py --dataset photo2cartoon --batch_size 4 --gpu_ids 0 1 2 3

Q&A

Q:为什么开源的卡通化模型与小程序中的效果有差异?

A:开源模型的训练数据收集自互联网,为了得到更加精美的效果,我们在训练小程序中卡通化模型时,采用了定制的卡通画数据(200多张),且增大了输入分辨率。此外,小程序中的人脸特征提取器采用自研的识别模型,效果优于本项目使用的开源识别模型。

Q:如何选取效果最好的模型?

A:首先训练模型200k iterations,然后使用FID指标挑选出最优模型,最终挑选出的模型为迭代90k iterations时的模型。

Q:关于人脸特征提取模型。

A:实验中我们发现,使用自研的识别模型计算Face ID Loss训练效果远好于使用开源识别模型,若训练效果出现鲁棒性问题,可尝试将Face ID Loss权重置零。

Q:人像分割模型是否能用与分割半身像?

A:不能。该模型是针对本项目训练的专用模型,需先裁剪出人脸区域再输入。

Tips

我们开源的模型是基于亚洲年轻女性训练的,对于其他人群覆盖不足,您可根据使用场景自行收集相应人群的数据进行训练。我们的开放平台提供了能够覆盖各类人群的卡通化服务,您可前往体验。如有定制卡通风格需求请联系商务:18852075216。

参考

U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation [Paper][Code]

InsightFace_Pytorch

Owner
Minivision_AI
Minivision_AI
Utilities to bridge Canvas-generated course rosters with GitLab's API.

gitlab-canvas-utils A collection of scripts originally written for CSE 13S. Oversees everything from GitLab course group creation, student repository

Eugene Chou 5 Jun 08, 2022
This package proposes simplified exporting pytorch models to ONNX and TensorRT, and also gives some base interface for model inference.

PyTorch Infer Utils This package proposes simplified exporting pytorch models to ONNX and TensorRT, and also gives some base interface for model infer

Alex Gorodnitskiy 11 Mar 20, 2022
Bio-OFC gym implementation and Gym-Fly environment

Bio-OFC gym implementation and Gym-Fly environment This repository includes the gym compatible implementation of the Bio-OFC algorithm from the paper

Siavash Golkar 1 Nov 16, 2021
Streamlit App For Product Analysis - Streamlit App For Product Analysis

Streamlit_App_For_Product_Analysis Здравствуйте! Перед вами дашборд, позволяющий

Grigory Sirotkin 1 Jan 10, 2022
Individual Tree Crown classification on WorldView-2 Images using Autoencoder -- Group 9 Weak learners - Final Project (Machine Learning 2020 Course)

Created by Olga Sutyrina, Sarah Elemili, Abduragim Shtanchaev and Artur Bille Individual Tree Crown classification on WorldView-2 Images using Autoenc

2 Dec 08, 2022
Adaout is a practical and flexible regularization method with high generalization and interpretability

Adaout Adaout is a practical and flexible regularization method with high generalization and interpretability. Requirements python 3.6 (Anaconda versi

lambett 1 Feb 09, 2022
Scalable, event-driven, deep-learning-friendly backtesting library

...Minimizing the mean square error on future experience. - Richard S. Sutton BTGym Scalable event-driven RL-friendly backtesting library. Build on

Andrew 922 Dec 27, 2022
Stochastic Normalizing Flows

Stochastic Normalizing Flows We introduce stochasticity in Boltzmann-generating flows. Normalizing flows are exact-probability generative models that

AI4Science group, FU Berlin (Frank Noé and co-workers) 50 Dec 16, 2022
PyTorch implementation of CloudWalk's recent work DenseBody

densebody_pytorch PyTorch implementation of CloudWalk's recent paper DenseBody. Note: For most recent updates, please check out the dev branch. Update

Lingbo Yang 401 Nov 19, 2022
The code for 'Deep Residual Fourier Transformation for Single Image Deblurring'

Deep Residual Fourier Transformation for Single Image Deblurring Xintian Mao, Yiming Liu, Wei Shen, Qingli Li and Yan Wang News 2021.12.5 Release Deep

145 Jan 05, 2023
Urban mobility simulations with Python3, RLlib (Deep Reinforcement Learning) and Mesa (Agent-based modeling)

Deep Reinforcement Learning for Smart Cities Documentation RLlib: https://docs.ray.io/en/master/rllib.html Mesa: https://mesa.readthedocs.io/en/stable

1 May 15, 2022
Unconstrained Text Detection with Box Supervisionand Dynamic Self-Training

SelfText Beyond Polygon: Unconstrained Text Detection with Box Supervisionand Dynamic Self-Training Introduction This is a PyTorch implementation of "

weijiawu 34 Nov 09, 2022
High-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently.

TL;DR Ignite is a high-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently. Click on the image to

4.2k Jan 01, 2023
Official repo for BMVC2021 paper ASFormer: Transformer for Action Segmentation

ASFormer: Transformer for Action Segmentation This repo provides training & inference code for BMVC 2021 paper: ASFormer: Transformer for Action Segme

42 Dec 23, 2022
Count the MACs / FLOPs of your PyTorch model.

THOP: PyTorch-OpCounter How to install pip install thop (now continously intergrated on Github actions) OR pip install --upgrade git+https://github.co

Ligeng Zhu 3.9k Dec 29, 2022
Paper: De-rendering Stylized Texts

Paper: De-rendering Stylized Texts Wataru Shimoda1, Daichi Haraguchi2, Seiichi Uchida2, Kota Yamaguchi1 1CyberAgent.Inc, 2 Kyushu University Accepted

CyberAgent AI Lab 55 Dec 18, 2022
Semi-supervised Transfer Learning for Image Rain Removal. In CVPR 2019.

Semi-supervised Transfer Learning for Image Rain Removal This package contains the Python implementation of "Semi-supervised Transfer Learning for Ima

Wei Wei 59 Dec 26, 2022
NeuroMorph: Unsupervised Shape Interpolation and Correspondence in One Go

NeuroMorph: Unsupervised Shape Interpolation and Correspondence in One Go This repository provides our implementation of the CVPR 2021 paper NeuroMorp

Meta Research 35 Dec 08, 2022
Improving Factual Consistency of Abstractive Text Summarization

Improving Factual Consistency of Abstractive Text Summarization We provide the code for the papers: "Entity-level Factual Consistency of Abstractive T

61 Nov 27, 2022
Localizing Visual Sounds the Hard Way

Localizing-Visual-Sounds-the-Hard-Way Code and Dataset for "Localizing Visual Sounds the Hard Way". The repo contains code and our pre-trained model.

Honglie Chen 58 Dec 07, 2022