File size: 1,364 Bytes
8bc6b47
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
import os
import gradio as gr
from PIL import Image

os.system("git clone https://github.com/autonomousvision/projected_gan")


os.chdir("projected_gan")

os.mkdir("outputs")

os.system("gdown --id '1H-MYFZqngF1R0whm4bc3fEoX7VvOWaDl'")

def inference(truncation,seeds):
  os.system("python gen_images.py --outdir=./outputs/ --trunc="+str(truncation)+" --seeds="+str(int(seeds))+" --network=network-snapshot-metfaces2.pkl")
  seeds = int(seeds)
  image = Image.open(f"./outputs/seed{seeds:04d}.png")
  return image

title = "Projected GAN"
description = "Gradio demo for Projected GAN. To use it, add seed and truncation, or click one of the examples to load them. Read more at the links below."

article = "<p style='text-align: center'><a href='http://www.cvlibs.net/publications/Sauer2021NEURIPS.pdf' target='_blank'>Projected GANs Converge Faster</a> | <a href='https://github.com/autonomousvision/projected_gan'>Github Repo</p><center><img src='https://visitor-badge.glitch.me/badge?page_id=akhaliq_projected_gan' alt='visitor badge'></center>"

gr.Interface(inference,[gr.inputs.Slider(label="truncation",minimum=0, maximum=5, step=0.1, default=0.8),gr.inputs.Slider(label="Seed",minimum=0, maximum=1000, step=1, default=0)],"pil",title=title,description=description,article=article, examples=[
    [0.8,0]
    ]).launch(enable_queue=True,cache_examples=True)