InterfaceGAN++: Exploring the limits of InterfaceGAN

Overview

InterfaceGAN++: Exploring the limits of InterfaceGAN

Authors: Apavou Clément & Belkada Younes

Python 3.8 pytorch 1.10.2 sklearn 0.21.2

Open In Colab

From left to right - Images generated using styleGAN and the boundaries Bald, Blond, Heavy_Makeup, Gray_Hair

This the the repository to a project related to the Introduction to Numerical Imaging (i.e, Introduction à l'Imagerie Numérique in French), given by the MVA Masters program at ENS-Paris Saclay. The project and repository is based on the work from Shen et al., and fully supports their codebase. You can refer to the original README) to reproduce their results.

Introduction

In this repository, we propose an approach, termed as InterFaceGAN++, for semantic face editing based on the work from Shen et al. Specifically, we leverage the ideas from the previous work, by applying the method for new face attributes, and also for StyleGAN3. We qualitatively explain that moving the latent vector toward the trained boundaries leads in many cases to keeping the semantic information of the generated images (by preserving its local structure) and modify the desired attribute, thus helps to demonstrate the disentangled property of the styleGANs.

🔥 Additional features

  • Supports StyleGAN2 & StyleGAN3 on the classic attributes
  • New attributes (Bald, Gray hair, Blond hair, Earings, ...) for:
    • StyleGAN
    • StyleGAN2
    • StyleGAN3
  • Supports face generation using StyleGAN3 & StyleGAN2

The list of new features can be found on our attributes detection classifier repository

🔨 Training an attribute detection classifier

We use a ViT-base model to train an attribute detection classifier, please refer to our classification code if you want to test it for new models. Once you retrieve the trained SVM from this repo, you can directly move them in this repo and use them.

Generate images using StyleGAN & StyleGAN2 & StyleGAN3

We did not changed anything to the structure of the old repository, please refer to the previous README. For StyleGAN

🎥 Get the pretrained StyleGAN

We use the styleGAN trained on ffhq for our experiments, if you want to reproduce them, run:

wget -P interfacegan/models/pretrain https://www.dropbox.com/s/qyv37eaobnow7fu/stylegan_ffhq.pth

🎥 Get the pretrained StyleGAN2

We use the styleGAN2 trained on ffhq for our experiments, if you want to reproduce them, run:

wget -P models/pretrain https://api.ngc.nvidia.com/v2/models/nvidia/research/stylegan2/versions/1/files/stylegan2-ffhq-1024x1024.pkl 

🎥 Get the pretrained StyleGAN3

We use the styleGAN3 trained on ffhq for our experiments, if you want to reproduce them, run:

wget -P models/pretrain https://api.ngc.nvidia.com/v2/models/nvidia/research/stylegan3/versions/1/files/stylegan3-t-ffhq-1024x1024.pkl 

The pretrained model should be copied at models/pretrain. If not, move the pretrained model file at this directory.

🎨 Run the generation script

If you want to generate 10 images using styleGAN3 downloaded before, run:

python generate_data.py -m stylegan3_ffhq -o output_stylegan3 -n 10

The arguments are exactly the same as the arguments from the original repository, the code supports the flag -m stylegan3_ffhq for styleGAN3 and -m stylegan3_ffhq for styleGAN2.

✏️ Edit generated images

You can edit the generated images using our trained boundaries! Depending on the generator you want to use, make sure that you have downloaded the right model and put them into models/pretrain.

Examples

Please refer to our interactive google colab notebook to play with our models by clicking the following badge:

Open In Colab

StyleGAN

Example of generated images using StyleGAN and moving the images towards the direction of the attribute grey hair:

original images generated with StyleGAN

grey hair version of the images generated with StyleGAN

StyleGAN2

Example of generated images using StyleGAN2 and moving the images towards the opposite direction of the attribute young:

original images generated with StyleGAN2

non young version of the images generated with StyleGAN2

StyleGAN3

Example of generated images using StyleGAN3 and moving the images towards the attribute beard:

Owner
Younes Belkada
MSc Student in Mathematics - Machine Learning - Perception | M2 MVA @ ENS Paris-Saclay
Younes Belkada
The Surprising Effectiveness of Visual Odometry Techniques for Embodied PointGoal Navigation

PointNav-VO The Surprising Effectiveness of Visual Odometry Techniques for Embodied PointGoal Navigation Project Page | Paper Table of Contents Setup

Xiaoming Zhao 41 Dec 15, 2022
Continuous Diffusion Graph Neural Network

We present Graph Neural Diffusion (GRAND) that approaches deep learning on graphs as a continuous diffusion process and treats Graph Neural Networks (GNNs) as discretisations of an underlying PDE.

Twitter Research 227 Jan 05, 2023
Cache Requests in Deta Bases and Echo them with Deta Micros

Deta Echo Cache Leverage the awesome Deta Micros and Deta Base to cache requests and echo them as needed. Stop worrying about slow public APIs or agre

Gingerbreadfork 8 Dec 07, 2021
Official PyTorch Code of GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Monocular 3D Object Detection (CVPR 2021)

GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Monocular 3D Object Detection GrooMeD-NMS: Grouped Mathematically Differentiable NMS for Mo

Abhinav Kumar 76 Jan 02, 2023
Image segmentation with private İstanbul Dataset

Image Segmentation This repo was created for academic research and test result. Repo will update after academic article online. This repo contains wei

İrem KÖMÜRCÜ 9 Dec 11, 2022
tensorflow code for inverse face rendering

InverseFaceRender This is tensorflow code for our project: Learning Inverse Rendering of Faces from Real-world Videos. (https://arxiv.org/abs/2003.120

Yuda Qiu 18 Nov 16, 2022
A generalist algorithm for cell and nucleus segmentation.

Cellpose | A generalist algorithm for cell and nucleus segmentation. Cellpose was written by Carsen Stringer and Marius Pachitariu. To learn about Cel

MouseLand 733 Dec 29, 2022
Course materials for Fall 2021 "CIS6930 Topics in Computing for Data Science" at New College of Florida

Fall 2021 CIS6930 Topics in Computing for Data Science This repository hosts course materials used for a 13-week course "CIS6930 Topics in Computing f

Yoshi Suhara 101 Nov 30, 2022
PyTorch implementation of Spiking Neural Networks trained on surrogate gradient & BPTT using snntorch.

snn-localization repo PyTorch implementation of Spiking Neural Networks trained on surrogate gradient & BPTT using snntorch. Install Dependencies Orig

Sami BARCHID 1 Jan 06, 2022
Running Google MoveNet Multipose Tracking models on OpenVINO.

MoveNet MultiPose Tracking on OpenVINO

60 Nov 17, 2022
PyTorch Implementation of VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis.

VAENAR-TTS - PyTorch Implementation PyTorch Implementation of VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis.

Keon Lee 67 Nov 14, 2022
Quasi-Dense Similarity Learning for Multiple Object Tracking, CVPR 2021 (Oral)

Quasi-Dense Tracking This is the offical implementation of paper Quasi-Dense Similarity Learning for Multiple Object Tracking. We present a trailer th

ETH VIS Research Group 327 Dec 27, 2022
Implementation of the paper "Fine-Tuning Transformers: Vocabulary Transfer"

Transformer-vocabulary-transfer Implementation of the paper "Fine-Tuning Transfo

LEYA 13 Nov 30, 2022
Multi-modal co-attention for drug-target interaction annotation and Its Application to SARS-CoV-2

CoaDTI Multi-modal co-attention for drug-target interaction annotation and Its Application to SARS-CoV-2 Abstract Environment The test was conducted i

Layne_Huang 7 Nov 14, 2022
Functional deep learning

Pipeline abstractions for deep learning. Full documentation here: https://lf1-io.github.io/padl/ PADL: is a pipeline builder for PyTorch. may be used

LF1 101 Nov 09, 2022
Extremely easy multi instancing software for minecraft speedrunning.

Easy Multi Extremely easy multi/single instancing software for minecraft speedrunning. A couple of goals of this project: Setup multi in minutes No fi

Duncan 8 Jul 16, 2022
QuanTaichi evaluation suite

QuanTaichi: A Compiler for Quantized Simulations (SIGGRAPH 2021) Yuanming Hu, Jiafeng Liu, Xuanda Yang, Mingkuan Xu, Ye Kuang, Weiwei Xu, Qiang Dai, W

Taichi Developers 120 Jan 04, 2023
Algorithmic trading using machine learning.

Algorithmic Trading This machine learning algorithm was built using Python 3 and scikit-learn with a Decision Tree Classifier. The program gathers sto

Sourav Biswas 101 Nov 10, 2022
maximal update parametrization (µP)

Maximal Update Parametrization (μP) and Hyperparameter Transfer (μTransfer) Paper link | Blog link In Tensor Programs V: Tuning Large Neural Networks

Microsoft 694 Jan 03, 2023
transfer attack; adversarial examples; black-box attack; unrestricted Adversarial Attacks on ImageNet; CVPR2021 天池黑盒竞赛

transfer_adv CVPR-2021 AIC-VI: unrestricted Adversarial Attacks on ImageNet CVPR2021 安全AI挑战者计划第六期赛道2:ImageNet无限制对抗攻击 介绍 : 深度神经网络已经在各种视觉识别问题上取得了最先进的性能。

25 Dec 08, 2022