MineGAN: effective knowledge transfer from GANs to target domains with few images

Abstract:

One of the attractive characteristics of deep neural networks is their ability to transfer knowledge obtained in one domain to other related domains. As a result, high-quality networks can be trained in domains with relatively little training data. This property has been extensively studied for discriminative networks but has received significantly less attention for generative models. Given the often enormous effort required to train GANs, both computationally as well as in the dataset collection, the re-use of pretrained GANs is a desirable objective. We propose a novel knowledge transfer method for generative models based on mining the knowledge that is most beneficial to a specific target domain, either from a single or multiple pretrained GANs. This is done using a miner network that identifies which part of the generative distribution of each pretrained GAN outputs samples closest to the target domain. Mining effectively steers GAN sampling towards suitable regions of the latent space, which facilitates the posterior finetuning and avoids pathologies of other methods such as mode collapse and lack of flexibility. We perform experiments on several complex datasets using various GAN architectures (BigGAN, Progressive GAN) and show that the proposed method, called MineGAN, effectively transfers knowledge to domains with few target images, outperforming existing methods. In addition, MineGAN can successfully transfer knowledge from multiple pretrained GANs.

Updating

Training for MineGAN on StyleGANv2

Training for MineGAN on StyleGAN

Training for MineGAN on mnist

Overview

• Dependences
• Installation
• Instructions
• Results
• References
• Contact

Dependences

• Python2.7, NumPy, SciPy, NVIDIA GPU
• Tensorflow/Pytorch: the version of tensorflow should be more 1.0(https://www.tensorflow.org/), pytorch is more 0.4
• Dataset: MNIST, CelebA, HHFQ, Imagenet, Places365 or your dataset

Installation

• Install tensorflow/pytorch

Instructions

• git clone [email protected]:yaxingwang/MineGAN.git to get MineGA

• Pretrained model: downloading the pretrained model from Biggan, and put it into data/your_data/weights. Note using G_ema.pth to replace G.pth, since we dones't use ema. The pretrained model is moved into MineGA/weights/biggan

• Downloading inception model and moving it into MineGA

• Preparing data: leveraging sh scripts/utils/prepare_data.py, and put it into data/your_data/data. Please check Biggan to learn how to generate the data

• Traing: sh scripts/launch_BigGAN_bs256x8.sh

References

• [1] BigGAN
• [2] PGAN

Contact

If you run into any problems with this code, please submit a bug report on the Github site of the project. For another inquries pleace contact with me: [email protected]