Skip to content

Instantly share code, notes, and snippets.

require 'torch'
require 'nn'
require 'image'
require 'optim'
require 'loadcaffe'
local cmd = torch.CmdLine()
@ProGamerGov
ProGamerGov / linear-color-transfer.py
Created February 16, 2017 17:41
This script performs the linear color transfer step that the Scale Control feature in leongatys/NeuralImageSynthesis performs.
#This script performs the linear color transfer step that
#leongatys/NeuralImageSynthesis' Scale Control code performs.
#https://github.com/leongatys/NeuralImageSynthesis/blob/master/ExampleNotebooks/ScaleControl.ipynb
import scipy
import h5py
import skimage
import os
from skimage import io,transform,img_as_float
from skimage.io import imread,imsave
from collections import OrderedDict
@ProGamerGov
ProGamerGov / neural_style_criterion.lua
Created March 2, 2017 23:15
Testing a new "criterion" parameter
require 'torch'
require 'nn'
require 'image'
require 'optim'
require 'loadcaffe'
local cmd = torch.CmdLine()
@ProGamerGov
ProGamerGov / neural_style_l2_dd_fc_labels_t7.lua
Last active June 11, 2017 22:33
New features include ".t7" Torch7 model support, Inception, ResNet, etc... architecture support, DeepDream, FC layers (caffemodels only) can be used as content layers and DeepDream layers, the ability to use label files, and an experimental pixel decay/L2 latent state regularizer.
-- Original DeepDream code by github.com/jcjohnson
-- Simaltaneous DeepDream and style transfer modifications by github.com/ProGamerGov
-- The FC layers as content layers feature, and the label file features come from:
-- github.com/htoyryla's gist.github.com/htoyryla/806ca4d978f0528114282cd00022ad71
-- Torch model and architecture support from https://github.com/szagoruyko/neural-style/tree/torch
require 'torch'
require 'nn'
@ProGamerGov
ProGamerGov / cuda_utils.cu
Last active June 11, 2017 23:22
Neural-Style with the laplacian feature from deep-photo-styletransfer. See the comments for setting up and using the new feature.
extern "C" {
#include "lua.h"
#include "lualib.h"
#include "lauxlib.h"
}
#include "luaT.h"
#include "THC.h"
#include <stdio.h>
@ProGamerGov
ProGamerGov / laplacian.py
Last active April 21, 2017 10:39
Working Laplacian Creator
# Standalone version of the Matting Laplacian code from here: https://github.com/martinbenson/deep-photo-styletransfer
# Usage: python3 laplacian.py -in_dir <directory> -lap_dir <directory> -width <value>
# Install the depdendencies with: pip3 install numpy scipy Pillow
# This script is intended for use with artistic style transfer neural networks, and Deep Photo Style Transfer.
import argparse
import glob
import os
import shutil
import multiprocessing
-- Original DeepDream code by github.com/jcjohnson
-- Simaltaneous DeepDream and style transfer modifications by github.com/ProGamerGov
-- The FC layers as content layers feature, and the label file features come from:
-- github.com/htoyryla's gist.github.com/htoyryla/806ca4d978f0528114282cd00022ad71
-- Indepdent layer weights from: https://github.com/htoyryla/neural-style/blob/master/neural_style.lua
-- Torch model and architecture support from https://github.com/szagoruyko/neural-style/tree/torch
-- Matting laplacian code from: github.com/martinbenson/deep-photo-styletransfer/
require 'torch'
require 'nn'
require 'image'
require 'optim'
require 'loadcaffe'
local cmd = torch.CmdLine()
-- Original DeepDream code by github.com/jcjohnson
-- Simaltaneous DeepDream and style transfer modifications by github.com/ProGamerGov
-- The FC layers as content layers feature, and the label file features come from:
-- github.com/htoyryla's gist.github.com/htoyryla/806ca4d978f0528114282cd00022ad71
-- Indepdent layer weights from: https://github.com/htoyryla/neural-style/blob/master/neural_style.lua
-- Torch model and architecture support from https://github.com/szagoruyko/neural-style/tree/torch
-- A script like this one is requires for making Torch models compatible: https://gist.github.com/szagoruyko/8828e09cc4687afd324d
@ProGamerGov
ProGamerGov / neural_style_seg.lua
Last active December 27, 2018 03:02
Working Multi-Region Spatial Control in Neural-Style. Also known as "masked style transfer", and "semantic/segmented style transfer".
-- Original mask related code from: https://github.com/martinbenson/deep-photo-styletransfer
-- Modified mask code by github.com/ProGamerGov
require 'torch'
require 'nn'
require 'image'
require 'optim'
require 'loadcaffe'