import streamlit as st import os # Set the title of the app import torch import torch.nn as nn import torchvision.transforms as transforms from torchvision.models import vgg19,vgg16 from PIL import Image import matplotlib.pyplot as plt import numpy as np import os import torch.nn.functional as F import pandas as pd import torch.optim as optim import neural_style import argparse import os import sys import time import re # from PIL import ImageFilter, ImageResampling import numpy as np import torch from torch.optim import Adam from torch.utils.data import DataLoader from torchvision import datasets from torchvision import transforms import torch.onnx import utils from transformer_net import TransformerNet from vgg import Vgg16 from torch.utils.data import Dataset, DataLoader st.set_page_config(layout='wide') st.title("Neural Style transfer ") device = torch.device("cuda" if torch.cuda.is_available() else "cpu") def stylize_onnx(content_image, args): """ Read ONNX model and run it using onnxruntime """ # assert not args["export_onnx"] import onnxruntime ort_session = onnxruntime.InferenceSession(args["model"]) def to_numpy(tensor): return ( tensor.detach().cpu().numpy() if tensor.requires_grad else tensor.cpu().numpy() ) ort_inputs = {ort_session.get_inputs()[0].name: to_numpy(content_image)} ort_outs = ort_session.run(None, ort_inputs) img_out_y = ort_outs[0] return torch.from_numpy(img_out_y) def stylize(device,args): content_image = utils.load_image(args['content_image'], scale=args['content_scale']) content_transform = transforms.Compose([ transforms.ToTensor(), transforms.Lambda(lambda x: x.mul(255)) ]) content_image = content_transform(content_image) content_image = content_image.unsqueeze(0).to(device) if args['model'].endswith(".onnx"): output = stylize_onnx(content_image, args) else: with torch.no_grad(): style_model = TransformerNet() state_dict = torch.load( args['model']) # remove saved deprecated running_* keys in InstanceNorm from the checkpoint for k in list(state_dict.keys()): if re.search(r'in\d+\.running_(mean|var)$', k): del state_dict[k] style_model.load_state_dict(state_dict) style_model.to(device) style_model.eval() if args['export_onnx'] != False: assert args['export_onnx_name'].endswith(".onnx"), "Export model file should end with .onnx" output = torch.onnx.export( style_model, content_image, args['export_onnx_name'], opset_version=11, ) else: output = style_model(content_image).cpu() utils.save_image(args['output_image'], output[0]) return True # args = { # "content_image":"hemant_content3.jpg", # "content_scale":0.95, # "model":"epoch_100_Fri_Apr_19_00_29_00_2024_10_100000.model", # "export_onnx": False, # "export_onnx_name":"natural_art_style.onnx", # "output_image":"./kavitha/hemanth_style13.jpg" # } # stylize(device,args) def run_model(args): # Function to run the model # This is where you can place your model execution code st.write("Running the model...") # Simulate model running (you can replace this with your actual model code) import time result = stylize(device,args) # Simulate model running for 2 seconds st.success("Model has run successfully!") if result: st.subheader('Output Image') width = 400 height = 400 st.image(args["output_image"], caption='Final output',width=width) # if result: # width = 400 # height = 400 # st.image(args["output_image"], caption='Final output',width=width) # def app(): # # Streamlit app title # args = { # } # # Allow user to upload an image file # uploaded_file = st.file_uploader('Upload a JPEG image', type=['jpg', 'jpeg']) # # Check if a file has been uploaded # if uploaded_file is not None: # # Verify the uploaded file's format (extension) # file_extension = uploaded_file.name.split('.')[-1].lower() # if file_extension in ['jpg', 'jpeg']: # # Specify the directory path where the image will be stored # directory_path = 'C:\\Users\\Kavitha padala\\Desktop\\HemanthDL\\examples\\fast_neural_style\\neural_style\\images\\content_images' # # Create the directory if it doesn't exist # if not os.path.exists(directory_path): # os.makedirs(directory_path) # file_name = "content_image" # # Generate the full path for the image file # file_path = os.path.join(directory_path, file_name+"."+ file_extension) # # Save the uploaded image to the specified directory # with open(file_path, 'wb') as f: # f.write(uploaded_file.getbuffer()) # model_sets = ['women_style_art_model_best1.model', 'women_style_art_model_best1.model'] # # Create a select box to allow users to choose one model set # selected_model_set = st.selectbox( # 'Choose a model set:', # Label for the select box # model_sets # List of model sets # ) # output_image_name = "style_image.jpg" # args["content_image"] = file_path # args["model"] = selected_model_set # args["content_scale"] = 0.95 # args["output_image"] = "C:\\Users\\Kavitha padala\\Desktop\\HemanthDL\\examples\\fast_neural_style\\neural_style\\images\\style_images\\" + output_image_name # args["export_onnx"] = False # if st.button('Run Model'): # run_model(args) # # Display the selected model set # st.write(f'You selected: {selected_model_set}') # else: # # If the file is not in the correct format, display an error message # st.error('Please upload a JPG image file.') # if __name__ == "__main__": # app() # def display_image(): # def run_style_transfer(content_image, style_image, model): # # Placeholder function to run the style transfer model # # Replace this function with your actual model execution code # st.write(f"Running model: {model}") # # Simulate processing (you can replace this with your actual model code) # import time # time.sleep(3) # # For demonstration, we'll just return the style image as the output # # (You would return the styled image as the final output in your actual model code) # return style_image def app(): # Streamlit app title # Create a layout with three columns left_col, middle_col, right_col = st.columns([1, 1, 2]) args = {} result = False; # Middle column: Display title # Left column: Ask user to upload an image with left_col: st.subheader('Upload Image') uploaded_image = st.file_uploader('Choose an image', type=['jpg', 'jpeg']) if uploaded_image is not None: # Verify the uploaded file's format (extension) file_extension = uploaded_image.name.split('.')[-1].lower() if file_extension in ['jpg', 'jpeg']: # Specify the directory path where the image will be stored directory_path = 'uploads' # Create the directory if it doesn't exist if not os.path.exists(directory_path): os.makedirs(directory_path) file_name = "content_image" # Generate the full path for the image file file_path = os.path.join(directory_path, file_name+"."+ file_extension) args["content_image"] = file_path # Save the uploaded image to the specified directory with open(file_path, 'wb') as f: f.write(uploaded_image.getbuffer()) st.image(uploaded_image,width=256) # Right column top: Allow user to select a model and display style image with middle_col: # Define a list of models style_images = ['Art Style', 'Women Art','glass art'] st.subheader('Style model') # Create a select box to choose a model selected_image = st.selectbox('Choose a model will be selected :', style_images) models_dict = { 'Art Style':'art_style.model', 'Women Art':'women_style.model', 'glass art':'glass_art.model' } output_image_name = "style_image.jpg" args["model"] = models_dict[selected_image] args["content_scale"] = 0.95 args["output_image"] = output_image_name args["export_onnx"] = False # Left column: After clicking "Run Model" button, display the final output style transfer image with right_col: if uploaded_image is not None and selected_image in models_dict: # Get the selected style image # style_image_path = style_images[selected_model] # # Load the uploaded image and style image # content_image = uploaded_image # style_image = style_image_path if st.button('Run Model'): result = run_model(args) # Create a button to run the model # if st.button('Run Model'): # # Run style transfer model # final_output = run_style_transfer(content_image, style_image, selected_model) # # Resize the final output image # # You can specify the desired width and height for the resized image # width = 400 # Replace with your desired width # height = 400 # Replace with your desired height # # Display the final output image with specified width and height # st.subheader('Final Output') # st.image(final_output, caption='Styled Image', width=width, height=height) if __name__ == "__main__": app()