Super-Resolution Google Colab + colab folder

colab-notebooks/README.md

@@ -0,0 +1 @@
+Google Colab Notebooks

colab-notebooks/Super_deep_daze.ipynb

@@ -0,0 +1,230 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "Super-deep-daze.ipynb",
+      "private_outputs": true,
+      "provenance": [],
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "yNomPmYphEf6"
+      },
+      "source": [
+        "# Super deep-daze\n",
+        "## deep-daze & Super-Resolution GAN\n",
+        "\n",
+        "## https://github.com/lucidrains/deep-daze\n",
+        "\n",
+        "## https://github.com/krasserm/super-resolution"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "G2PhQt5gi2ag"
+      },
+      "source": [
+        "# deep-daze"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "Vy_VWKIFhB34"
+      },
+      "source": [
+        "**Restart after running this cell!**\n",
+        "\n",
+        "You must run this cell and then restart and rerun everything for the PyTorch version to be correct. Otherwise the model will run but not produce any meaningful output."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "qs1XUMEYg7B_"
+      },
+      "source": [
+        "!nvidia-smi\n",
+        "\n",
+        "import subprocess\n",
+        "\n",
+        "CUDA_version = [s for s in subprocess.check_output([\"nvcc\", \"--version\"]).decode(\"UTF-8\").split(\", \") if s.startswith(\"release\")][0].split(\" \")[-1]\n",
+        "print(\"CUDA version:\", CUDA_version)\n",
+        "\n",
+        "if CUDA_version == \"10.0\":\n",
+        "    torch_version_suffix = \"+cu100\"\n",
+        "elif CUDA_version == \"10.1\":\n",
+        "    torch_version_suffix = \"+cu101\"\n",
+        "elif CUDA_version == \"10.2\":\n",
+        "    torch_version_suffix = \"\"\n",
+        "else:\n",
+        "    torch_version_suffix = \"+cu110\"\n",
+        "\n",
+        "!pip install torch==1.7.1{torch_version_suffix} torchvision==0.8.2{torch_version_suffix} -f https://download.pytorch.org/whl/torch_stable.html ftfy regex\n",
+        "!pip install deep-daze --upgrade"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "cellView": "form",
+        "id": "sl-7OtJHhVvF"
+      },
+      "source": [
+        "from tqdm import trange\n",
+        "from IPython.display import Image, display\n",
+        "import os\n",
+        "from deep_daze import Imagine\n",
+        "\n",
+        "TEXT = \"Super Deep Daze\" #@param {type:\"string\"}\n",
+        "NUM_LAYERS = 42 #@param {type:\"number\"}\n",
+        "SAVE_EVERY =  10#@param {type:\"number\"}\n",
+        "IMAGE_WIDTH = 512 #@param {type:\"number\"}\n",
+        "SAVE_PROGRESS = True #@param {type:\"boolean\"}\n",
+        "LEARNING_RATE = 1e-5 #@param {type:\"number\"}\n",
+        "ITERATIONS = 10000 #@param {type:\"number\"}\n",
+        "\n",
+        "os.chdir('/content/')\n",
+        "\n",
+        "model = Imagine(\n",
+        "                text = TEXT,\n",
+        "                num_layers = NUM_LAYERS,\n",
+        "                save_every = SAVE_EVERY,\n",
+        "                image_width = IMAGE_WIDTH,\n",
+        "                batch_size=64,\n",
+        "                gradient_accumulate_every=1,\n",
+        "                lr = LEARNING_RATE,\n",
+        "                iterations = ITERATIONS,\n",
+        "                save_progress = SAVE_PROGRESS,\n",
+        "                save_date_time = True).cuda()\n",
+        "\n",
+        "for epoch in trange(20, desc = 'epochs'):\n",
+        "    for i in trange(ITERATIONS, desc = 'iteration'):\n",
+        "        model.train_step(epoch, i)\n",
+        "\n",
+        "        if i % model.save_every != 0:\n",
+        "            continue\n",
+        "\n",
+        "        filename = TEXT.replace(' ', '_')\n",
+        "        image = Image(filename + '.png')\n",
+        "        display(image)"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "Ny7IiA3PicJK"
+      },
+      "source": [
+        "# Super-Resolution GAN"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "HKMvpfahiXjR"
+      },
+      "source": [
+        "%cd /content/\n",
+        "!git clone https://github.com/krasserm/super-resolution\n",
+        "%cd /content/super-resolution/"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "kiYb41o1if7I"
+      },
+      "source": [
+        "import os\n",
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "from data import DIV2K\n",
+        "from model.srgan import generator, discriminator\n",
+        "\n",
+        "from model import resolve_single\n",
+        "from utils import load_image\n",
+        "\n",
+        "%matplotlib inline"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "tA20tQQvihmO"
+      },
+      "source": [
+        "# Location of model weights (needed for demo)\n",
+        "weights_dir = 'weights/srgan'\n",
+        "weights_file = lambda filename: os.path.join(weights_dir, filename)\n",
+        "\n",
+        "os.makedirs(weights_dir, exist_ok=True)"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "vZfxW83Uimoc"
+      },
+      "source": [
+        "%cd /content/super-resolution/\n",
+        "!wget https://martin-krasser.de/sisr/weights-srgan.tar.gz\n",
+        "!tar -xpf weights-srgan.tar.gz\n",
+        "%cd /content/super-resolution"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "nHILuxRkippo"
+      },
+      "source": [
+        "gan_generator = generator()\n",
+        "gan_generator.load_weights(weights_file('gan_generator.h5'))\n",
+        "\n",
+        "lr = load_image('/content/Super_Deep_Daze.jpg')\n",
+        "image = resolve_single(gan_generator, lr)"
+      ],
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "YHxF1BLAiwZ0"
+      },
+      "source": [
+        "plt.figimage(image, resize=True)\n",
+        "#plt.savefig('/content/TEST')\n",
+        "plt.imshow(image)"
+      ],
+      "execution_count": null,
+      "outputs": []
+    }
+  ]
+}

deep_daze/deep_daze.py

@@ -267,7 +267,7 @@ def replace_current_img(self):
         if os.path.isfile(always_current_img) or os.path.islink(always_current_img):
             os.remove(always_current_img)  # remove the file
 
-        copy(str(self.filename), always_current_img)
+        #copy(str(self.filename), str(self.filename))
 
     def generate_and_save_image(self, custom_filename: Path = None, current_iteration: int = None):
         """
@@ -279,7 +279,7 @@ def generate_and_save_image(self, custom_filename: Path = None, current_iteratio
             img.clamp_(0., 1.)
             self.filename = custom_filename if custom_filename else self.image_output_path(current_iteration=current_iteration)
             save_image(img, self.filename)
-            self.replace_current_image()
+            self.replace_current_img()
             tqdm.write(f'image updated at "./{str(self.filename)}"')