app.py

import streamlit as st
import urllib.request
import PIL.Image
from PIL import Image
import requests
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
from urllib.request import urlretrieve
from io import BytesIO
import numpy as np
import torchvision.transforms as T
from PIL import Image,ImageOps,ImageFilter
from io import BytesIO
import os




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 getNeighbours(i, j, n, m) :
    arr = []
    if i-1 >= 0 and j-1 >= 0 :
        arr.append((i-1, j-1))
    if i-1 >= 0 :
        arr.append((i-1, j))
    if i-1 >= 0 and j+1 < m :
        arr.append((i-1, j+1))
    if j+1 < m :
        arr.append((i, j+1))
    if i+1 < n and j+1 < m :
        arr.append((i+1, j+1))
    if i+1 < n :
        arr.append((i+1, j))
    if i+1 < n and j-1 >= 0 :
        arr.append((i+1, j-1))
    if j-1 >= 0 :
        arr.append((i, j-1))
    return arr

MODEL_URL = "https://www.dropbox.com/s/05ong36r29h51ov/popd.pkl?dl=1"
urllib.request.urlretrieve(MODEL_URL, "popd.pkl")
path = Path(".")
learn=load_learner(path, 'popd.pkl')

def predict(image,colour):
  img_fast = open_image(image)
  a = PIL.Image.open(image).convert('RGB')
  st.image(a, caption='Input')
  p,img_hr,b = learn.predict(img_fast)
  x = np.minimum(np.maximum(image2np(img_hr.data*255), 0), 255).astype(np.uint8)
  img = PIL.Image.fromarray(x).convert('RGB')
  size = a.size
  im1 = img.resize(size)
  membuf = BytesIO()
  im1.save(membuf, format="png") 
  im = Image.open(membuf)
  im = im.convert('RGBA')
  data = np.array(im)   # "data" is a height x width x 4 numpy array
  red, green, blue, alpha = data.T # Temporarily unpack the bands for readability'
  white_areas = (red == 0) & (blue == 0) & (green == 0)
  data[..., :-1][white_areas.T] = (0,0,0) # Transpose back needed
  im2 = Image.fromarray(data)
  membuf = BytesIO()
  im2.save(membuf, format="png") 
  img = Image.open(membuf)
  bitmap = img.load()
  n = img.size[0]
  m = img.size[1]
  stateMap = []
  for i in range(n):
    stateMap.append([False for j in range(m)])  
  queue = [(0, 0)]
  while queue:
    e = queue.pop(0)
    i = e[0]
    j = e[1]
    if not stateMap[i][j]:
      stateMap[i][j] = True
      color = int((bitmap[i, j][0] + bitmap[i, j][1] + bitmap[i, j][2])/3)
      if color > 100:
        bitmap[i, j] =colour
        neigh = getNeighbours(i, j, n, m)
        for ne in neigh:
          queue.append(ne)

  return st.image(img, caption='PoP ArT')
 
SIDEBAR_OPTION_DEMO_IMAGE = "Select a Demo Image"
SIDEBAR_OPTION_UPLOAD_IMAGE = "Upload an Image"
#SIDEBAR_OPTION_COLOUR_IMAGE = "Choose a colour"

SIDEBAR_OPTIONS = [SIDEBAR_OPTION_DEMO_IMAGE, SIDEBAR_OPTION_UPLOAD_IMAGE]
st.sidebar.write("Check out GitHub [link](https://github.com/vijishmadhavan/PoPd)")


app_mode = st.sidebar.selectbox("Please select from the following", SIDEBAR_OPTIONS)
photos = ["fight.jpg","shaolin-kung-fu.jpg","unnamed.jpg","michael-jackson.png"]
colour = ['Red','Blue','Yellow']
if app_mode == SIDEBAR_OPTION_DEMO_IMAGE:
  st.sidebar.write(" ------ ")
  option = st.sidebar.selectbox('Please select a sample image,colour and then click PoP button', photos)
  colour = st.sidebar.selectbox("Colour", colour)
  if colour == 'Red':
    colour = (185, 39, 40)
  elif colour == 'Blue':
    colour = (40, 96, 219)
  else:
    colour = (249, 223, 2)
  pressed = st.sidebar.button('PoP')
  if pressed:
    st.empty()
    st.sidebar.write('Please wait for the magic to happen! This may take up to a minute.')
    predict(option,colour)

elif app_mode == SIDEBAR_OPTION_UPLOAD_IMAGE:
  uploaded_file = st.file_uploader("Choose an image...")
  if uploaded_file is not None:
    colour = st.sidebar.selectbox("Colour", colour)
    if colour == 'Red':
      colour = (185, 39, 40)
    elif colour == 'Blue':
      colour = (40, 96, 219)
    else:
      colour = (249, 223, 2)
    pressed = st.sidebar.button('PoP')
    if pressed:
      st.empty()
      st.sidebar.write('Please wait for the magic to happen! This may take up to a minute.')
      predict(uploaded_file,colour)