app.py

import fastai
from fastai.vision import *
from fastai.utils import *
from fastai.vision import open_image, load_learner, image, torch
import numpy as np
import urllib.request
import PIL.Image
from io import BytesIO
import torchvision.transforms as T
from PIL import Image
import requests
from io import BytesIO
import fastai
from fastai.vision import *
from fastai.utils.mem import *
from fastai.vision import open_image, load_learner, image, torch
import numpy as np
import urllib.request
import PIL.Image
from PIL import Image
from io import BytesIO
import torchvision.transforms as T

class FeatureLoss(nn.Module):
    def __init__(self, m_feat, layer_ids, layer_wgts):
        super().__init__()
        self.m_feat = m_feat
        self.loss_features = [self.m_feat[i] for i in layer_ids]
        self.hooks = hook_outputs(self.loss_features, detach=False)
        self.wgts = layer_wgts
        self.metric_names = ['pixel',] + [f'feat_{i}' for i in range(len(layer_ids))
              ] + [f'gram_{i}' for i in range(len(layer_ids))]

    def make_features(self, x, clone=False):
        self.m_feat(x)
        return [(o.clone() if clone else o) for o in self.hooks.stored]
    
    def forward(self, input, target):
        out_feat = self.make_features(target, clone=True)
        in_feat = self.make_features(input)
        self.feat_losses = [base_loss(input,target)]
        self.feat_losses += [base_loss(f_in, f_out)*w
                             for f_in, f_out, w in zip(in_feat, out_feat, self.wgts)]
        self.feat_losses += [base_loss(gram_matrix(f_in), gram_matrix(f_out))*w**2 * 5e3
                             for f_in, f_out, w in zip(in_feat, out_feat, self.wgts)]
        self.metrics = dict(zip(self.metric_names, self.feat_losses))
        return sum(self.feat_losses)
    
    def __del__(self): self.hooks.remove()

def add_margin(pil_img, top, right, bottom, left, color):
    width, height = pil_img.size
    new_width = width + right + left
    new_height = height + top + bottom
    result = Image.new(pil_img.mode, (new_width, new_height), color)
    result.paste(pil_img, (left, top))
    return result


MODEL_URL = "https://www.dropbox.com/s/04suaimdpru76h3/ArtLine_920.pkl?dl=1 "
urllib.request.urlretrieve(MODEL_URL, "ArtLine_920.pkl")
path = Path(".")
print(os.listdir('.'))
learn=load_learner(path, 'ArtLine_920.pkl')


import gradio as gr
import cv2


def get_filename(prefix="sketch"):
  from datetime import datetime
  from pytz import timezone
  return datetime.now(timezone('Asia/Seoul')).strftime('sketch__%Y-%m-%d %H:%M:%S.jpg')

def predict(img):
  img = PIL.Image.fromarray(img)
  img = add_margin(img, 250, 250, 250, 250, (255, 255, 255))
  img = np.array(img)

  h, w = img.shape[:-1]
  cv2.imwrite("test.jpg", img)
  img_test = open_image("test.jpg")

  p,img_hr,b = learn.predict(img_test)

  res = (img_hr / img_hr.max()).numpy()
  res = res[0]  # take only first channel as result
  res = cv2.resize(res, (w,h))

  output_file = get_filename()

  cv2.imwrite(output_file, (res * 255).astype(np.uint8), [cv2.IMWRITE_JPEG_QUALITY, 50])

  return res, output_file

gr.Interface(predict, 
             inputs="image", 
             outputs=[gr.Image(label="Sketch Image"), gr.File(label="Result File")],
             title="Image-to-sketch",
             description="Transfer any image into BW cartoon-style sketch!").launch()