ProGamerGov

neural-style-pt with histogram loss

The code here is based on genekogan's neural-style-pt histogram loss code. The CUDA code comes from pierre-wilmot's code here: https://github.com/pierre-wilmot/NeuralTextureSynthesis

Histogram Loss Layers

Each histogram loss layer stores the style image's histogram as a target, and then uses that compute the difference to the image being stylized.

Setup

neural-style-pt with mean loss and histogram transfer

Histogram Transfer

Users can specify an image for which the histogram will be transfered from, and what images the histogram will be transfered to; either the content image, style image(s), or both.

Mean Loss

A new loss layer type has been added that uses image means. Currently it only uses the first style image specified.

neural-style-pt with tiling

Tiling

The tiling feature is based on neural-dream's tiling system.

Usage

Basic usage:

python neural_style_tile.py -style_image  -content_image  -tile_size 256 -image_size 512

	#!/bin/bash
	fileid="1qWjJd-mVNxvHhzc9XsU6LjC0pQXIwW4t"
	filename="models/VGG16-Stylized-ImageNet.pth"
	curl -c ./cookie -s -L "https://drive.google.com/uc?export=download&id=${fileid}" > /dev/null
	curl -Lb ./cookie "https://drive.google.com/uc?export=download&confirm=`awk '/download/ {print $NF}' ./cookie`&id=${fileid}" -o ${filename}

	fileid="1EU-F9ugeIeTO9ay4PinzsBXgEuCYBu0Z"
	filename="models/VGG16_SOD_finetune.pth"
	curl -c ./cookie -s -L "https://drive.google.com/uc?export=download&id=${fileid}" > /dev/null
	curl -Lb ./cookie "https://drive.google.com/uc?export=download&confirm=`awk '/download/ {print $NF}' ./cookie`&id=${fileid}" -o ${filename}

	import os
	import copy
	import torch
	import torch.nn as nn
	import torch.optim as optim
	import torchvision.transforms as transforms

	from PIL import Image
	from CaffeLoader import loadCaffemodel, ModelParallel

	import os
	import copy
	import torch
	import torch.nn as nn
	import torch.optim as optim
	import torchvision.transforms as transforms

	from PIL import Image
	from CaffeLoader import loadCaffemodel, ModelParallel

	import torch
	import torch.nn as nn
	from collections import OrderedDict
	import imp
	import numpy as np


	# Import the model classes that were edited. Replace 'model_class_name" with the name of the class script, and
	# replace 'ModelName' with the name of the class in the script
	from model_class_name import ModelName

	# tensorflow/lucid CPPN (X,Y) --> (R,G,B) Differentiable Image Parameterization in PyTorch
	import torch
	import torch.nn as nn
	import torch.optim as optim
	import torchvision.transforms as transforms
	from inception5h import Inception5h
	from PIL import Image

	from typing import Tuple


	import torch


	def color_transfer(
	input: torch.Tensor,
	source: torch.Tensor,
	mode: str = "pca",

	from collections import OrderedDict
	from typing import Callable, Dict, Optional
	from warnings import warn

	import torch

	def _remove_all_forward_hooks(
	module: torch.nn.Module, hook_fn_name: Optional[str] = None
	) -> None:
	"""