File size: 12,129 Bytes
ed697ed |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 |
{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "0aWBuYb2UDIO"
},
"source": [
"# SAM: Animation Inference Playground"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "Uuviq3qQkUFy"
},
"outputs": [],
"source": [
"import os\n",
"os.chdir('/content')\n",
"CODE_DIR = 'SAM'"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "QQ6XEmlHlXbk",
"outputId": "2d2af9bb-1bbe-4946-84db-dbeaf62aa226"
},
"outputs": [],
"source": [
"!git clone https://github.com/yuval-alaluf/SAM.git $CODE_DIR"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "JaRUFuVHkzye",
"outputId": "12fc6dcd-951b-472f-b9d6-0a09f749e931"
},
"outputs": [],
"source": [
"!wget https://github.com/ninja-build/ninja/releases/download/v1.8.2/ninja-linux.zip\n",
"!sudo unzip ninja-linux.zip -d /usr/local/bin/\n",
"!sudo update-alternatives --install /usr/bin/ninja ninja /usr/local/bin/ninja 1 --force "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "23baccYQlU9E"
},
"outputs": [],
"source": [
"os.chdir(f'./{CODE_DIR}')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "lEWIzkaLSsFY"
},
"outputs": [],
"source": [
"from argparse import Namespace\n",
"import os\n",
"import sys\n",
"import pprint\n",
"import numpy as np\n",
"from PIL import Image\n",
"import torch\n",
"import torchvision.transforms as transforms\n",
"\n",
"sys.path.append(\".\")\n",
"sys.path.append(\"..\")\n",
"\n",
"from datasets.augmentations import AgeTransformer\n",
"from utils.common import tensor2im\n",
"from models.psp import pSp"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "9J3NEVlESsFl"
},
"outputs": [],
"source": [
"EXPERIMENT_TYPE = 'ffhq_aging'"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "eFjfO9q9SsFm"
},
"source": [
"## Step 1: Download Pretrained Model\n",
"As part of this repository, we provide our pretrained aging model.\n",
"We'll download the model for the selected experiments as save it to the folder `../pretrained_models`."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "e2L9GRCFSsFm"
},
"outputs": [],
"source": [
"def get_download_model_command(file_id, file_name):\n",
" \"\"\" Get wget download command for downloading the desired model and save to directory ../pretrained_models. \"\"\"\n",
" current_directory = os.getcwd()\n",
" save_path = os.path.join(os.path.dirname(current_directory), \"pretrained_models\")\n",
" if not os.path.exists(save_path):\n",
" os.makedirs(save_path)\n",
" url = r\"\"\"wget --load-cookies /tmp/cookies.txt \"https://docs.google.com/uc?export=download&confirm=$(wget --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://docs.google.com/uc?export=download&id={FILE_ID}' -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\\1\\n/p')&id={FILE_ID}\" -O {SAVE_PATH}/{FILE_NAME} && rm -rf /tmp/cookies.txt\"\"\".format(FILE_ID=file_id, FILE_NAME=file_name, SAVE_PATH=save_path)\n",
" return url"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "yotXNk8PSsFn"
},
"outputs": [],
"source": [
"MODEL_PATHS = {\n",
" \"ffhq_aging\": {\"id\": \"1XyumF6_fdAxFmxpFcmPf-q84LU_22EMC\", \"name\": \"sam_ffhq_aging.pt\"}\n",
"}\n",
"\n",
"path = MODEL_PATHS[EXPERIMENT_TYPE]\n",
"download_command = get_download_model_command(file_id=path[\"id\"], file_name=path[\"name\"])"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "tXUqcxd8SsFo",
"outputId": "0e2426ec-f96f-44cb-d1a3-736807bc4f37"
},
"outputs": [],
"source": [
"!wget {download_command}"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "mismxuEvSsFp"
},
"source": [
"## Step 3: Define Inference Parameters"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "NRzP-untSsFq"
},
"source": [
"Below we have a dictionary defining parameters such as the path to the pretrained model to use and the path to the\n",
"image to perform inference on.\n",
"While we provide default values to run this script, feel free to change as needed."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "pHeJRfsnSsFq"
},
"outputs": [],
"source": [
"EXPERIMENT_DATA_ARGS = {\n",
" \"ffhq_aging\": {\n",
" \"model_path\": \"../pretrained_models/sam_ffhq_aging.pt\",\n",
" \"transform\": transforms.Compose([\n",
" transforms.Resize((256, 256)),\n",
" transforms.ToTensor(),\n",
" transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])\n",
" }\n",
"}"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "IgzLA96mSsFq"
},
"outputs": [],
"source": [
"EXPERIMENT_ARGS = EXPERIMENT_DATA_ARGS[EXPERIMENT_TYPE]"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "kqlL9U5uSsFr"
},
"source": [
"## Step 4: Load Pretrained Model\n",
"We assume that you have downloaded the pretrained aging model and placed it in the path defined above."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "8khx-2fcSsFr"
},
"outputs": [],
"source": [
"model_path = EXPERIMENT_ARGS['model_path']\n",
"ckpt = torch.load(model_path, map_location='cpu')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "fVfLlsfjSsFr",
"outputId": "ccc5cc29-e59d-414c-a216-fa967ece4eb9"
},
"outputs": [],
"source": [
"opts = ckpt['opts']\n",
"pprint.pprint(opts)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "vxYxSJk1SsFs"
},
"outputs": [],
"source": [
"# update the training options\n",
"opts['checkpoint_path'] = model_path"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "Z4LNoYt_SsFs",
"outputId": "030da871-e4a2-42a9-9c4b-8b96e028fd40"
},
"outputs": [],
"source": [
"opts = Namespace(**opts)\n",
"net = pSp(opts)\n",
"net.eval()\n",
"net.cuda()\n",
"print('Model successfully loaded!')"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "3mNH95EJSsFs"
},
"source": [
"### Utils for Generating MP4 "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"id": "GASwMVmJSsFs"
},
"outputs": [],
"source": [
"import imageio\n",
"from tqdm import tqdm\n",
"import matplotlib\n",
"from IPython.display import HTML\n",
"from base64 import b64encode\n",
"\n",
"matplotlib.use('module://ipykernel.pylab.backend_inline')\n",
"%matplotlib inline\n",
"\n",
"\n",
"def generate_mp4(out_name, images, kwargs):\n",
" writer = imageio.get_writer(out_name + '.mp4', **kwargs)\n",
" for image in images:\n",
" writer.append_data(image)\n",
" writer.close()\n",
"\n",
"\n",
"def run_on_batch_to_vecs(inputs, net):\n",
" _, result_batch = net(inputs.to(\"cuda\").float(), return_latents=True, randomize_noise=False, resize=False)\n",
" return result_batch.cpu()\n",
"\n",
"\n",
"def get_result_from_vecs(vectors_a, vectors_b, alpha):\n",
" results = []\n",
" for i in range(len(vectors_a)):\n",
" cur_vec = vectors_b[i] * alpha + vectors_a[i] * (1 - alpha)\n",
" res = net(cur_vec.cuda(), randomize_noise=False, input_code=True, input_is_full=True, resize=False)\n",
" results.append(res[0])\n",
" return results\n",
"\n",
"\n",
"def show_mp4(filename, width=400):\n",
" mp4 = open(filename + '.mp4', 'rb').read()\n",
" data_url = \"data:video/mp4;base64,\" + b64encode(mp4).decode()\n",
" display(HTML(\"\"\"\n",
" <video width=\"%d\" controls autoplay loop>\n",
" <source src=\"%s\" type=\"video/mp4\">\n",
" </video>\n",
" \"\"\" % (width, data_url)))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "y4oqFBfTSsFt",
"outputId": "6ff2e285-e5e0-4d9c-e16d-af43eee1e8f1"
},
"outputs": [],
"source": [
"SEED = 42\n",
"np.random.seed(SEED)\n",
"\n",
"img_transforms = EXPERIMENT_ARGS['transform']\n",
"n_transition = 25\n",
"kwargs = {'fps': 40}\n",
"save_path = \"notebooks/animations\"\n",
"os.makedirs(save_path, exist_ok=True)\n",
"\n",
"#################################################################\n",
"# TODO: define your image paths here to be fed into the model\n",
"#################################################################\n",
"root_dir = 'notebooks/images'\n",
"ims = ['866', '1287', '2468']\n",
"im_paths = [os.path.join(root_dir, im) + '.jpg' for im in ims]\n",
"\n",
"# NOTE: Please make sure the images are pre-aligned!\n",
"\n",
"target_ages = [0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 90, 80, 70, 60, 50, 40, 30, 20, 10, 0]\n",
"age_transformers = [AgeTransformer(target_age=age) for age in target_ages]\n",
"\n",
"for image_path in im_paths:\n",
" image_name = os.path.basename(image_path)\n",
" print(f'Working on image: {image_name}')\n",
" original_image = Image.open(image_path).convert(\"RGB\")\n",
" input_image = img_transforms(original_image)\n",
" all_vecs = []\n",
" for idx, age_transformer in enumerate(age_transformers):\n",
"\n",
" input_age_batch = [age_transformer(input_image.cpu()).to('cuda')]\n",
" input_age_batch = torch.stack(input_age_batch)\n",
"\n",
" # get latent vector for the current target age amount\n",
" with torch.no_grad():\n",
" result_vec = run_on_batch_to_vecs(input_age_batch, net)\n",
" result_image = get_result_from_vecs([result_vec], result_vec, 0)[0]\n",
" all_vecs.append([result_vec])\n",
"\n",
" images = []\n",
" for i in range(1, len(target_ages)):\n",
" alpha_vals = np.linspace(0, 1, n_transition).tolist()\n",
" for alpha in tqdm(alpha_vals):\n",
" result_image = get_result_from_vecs(all_vecs[i-1], all_vecs[i], alpha)[0]\n",
" output_im = tensor2im(result_image)\n",
" images.append(np.array(output_im))\n",
"\n",
" animation_path = os.path.join(save_path, f\"{image_name}_animation\")\n",
" generate_mp4(animation_path, images, kwargs)\n",
" show_mp4(animation_path)"
]
}
],
"metadata": {
"accelerator": "GPU",
"colab": {
"name": "inference_playground_mp4.ipynb",
"provenance": [],
"toc_visible": true
},
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.3"
}
},
"nbformat": 4,
"nbformat_minor": 1
} |