Skip to content

Latest commit

 

History

History

anemigan

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
cpp
cpp
 
 
python
python
 
 
README.md
README.md
 
 
README_CN.md
README_CN.md
 
 

README.md

English | 简体中文

Image Generation Model

Now FastDeploy supports the deployment of the following style transfer models in the PaddleHub pre-trained model repository

Model Description Model Format
animegan_v1_hayao_60 Convert the input image into one in Miyazaki anime style with model weights converting from AnimeGAN V1 official open source project paddle
animegan_v2_paprika_97 Convert the input image into one in Satoshi Paprika anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_hayao_64 Convert the input image into one in Miyazaki anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_shinkai_53 Convert the input image into one in Shinkai anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_shinkai_33 Convert the input image into one in Shinkai anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_paprika_54 Convert the input image into one in Satoshi Paprika anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_hayao_99 Convert the input image into one in Miyazaki anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_paprika_74 Convert the input image into one in Satoshi Paprika anime style with model weights converting from AnimeGAN V2 official open source project paddle
animegan_v2_paprika_98 Convert the input image into one in Satoshi Paprika anime style with model weights converting from AnimeGAN V2 official open source project paddle

Speed comparison between hub and FastDeploy paddle backend Deployment (ips, higher is better)

Device FastDeploy Hub
CPU 0.075 0.069
GPU 8.33 8.26

Downloading pre-trained models

Use fastdeploy.download_model to download models. For example, download animegan_v1_hayao_60

import fastdeploy as fd
fd.download_model(name='animegan_v1_hayao_60', path='./', format='paddle')

The pre-trained model of animegan_v1_hayao_60 will be available in the current directory.

Detailed deployment Tutorials