├── Experiments_CelebA
    ├── CelebA.ipynb
    ├── README.md
    ├── celeba_low_1000.pkl
    └── dataset_celebA.7z
├── Experiments_DeepFakeDetection
    ├── .gitkeep
    ├── FaceForensic.ipynb
    ├── README.md
    ├── radialProfile.py
    ├── test_1000.pkl
    └── train_3200.pkl
├── Experiments_Faces-HQ
    ├── .gitkeep
    ├── Faces-HQ.ipynb
    ├── README.md
    ├── Visualization.ipynb
    ├── dataset_freq_1000.pkl
    └── radialProfile.py
├── README.md
└── imgs
    ├── .gitkeep
    ├── 1000_celeba.png
    ├── 1000_deep.png
    ├── 1000_hq.png
    ├── celeba_results.PNG
    ├── dataset.png
    ├── deep_results.PNG
    ├── faces_results.PNG
    ├── pipeline.png
    └── results3.png


/Experiments_CelebA/CelebA.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# CelebA"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "markdown",
 12 |    "metadata": {},
 13 |    "source": [
 14 |     "In this notebook we show the results for CelebA. You can create from scratch the features or use the pre-computed ones."
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "markdown",
 19 |    "metadata": {},
 20 |    "source": [
 21 |     "### 1. Create feature"
 22 |    ]
 23 |   },
 24 |   {
 25 |    "cell_type": "markdown",
 26 |    "metadata": {},
 27 |    "source": [
 28 |     "If you want to create the features, first of all unzip file \"dataset_celebA.7z \". Be sure to save the folder together with this notebook. You also need to download  [CelebA](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html).\n",
 29 |     "\n",
 30 |     "Otherwise, just jump to section 2."
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 1,
 36 |    "metadata": {},
 37 |    "outputs": [
 38 |     {
 39 |      "name": "stdout",
 40 |      "output_type": "stream",
 41 |      "text": [
 42 |       "DATA Saved\n"
 43 |      ]
 44 |     }
 45 |    ],
 46 |    "source": [
 47 |     "import cv2\n",
 48 |     "import numpy as np\n",
 49 |     "import os\n",
 50 |     "import radialProfile\n",
 51 |     "import glob\n",
 52 |     "from matplotlib import pyplot as plt\n",
 53 |     "import pickle\n",
 54 |     "from scipy.interpolate import griddata\n",
 55 |     "\n",
 56 |     "data= {}\n",
 57 |     "epsilon = 1e-8\n",
 58 |     "N = 80\n",
 59 |     "y = []\n",
 60 |     "error = []\n",
 61 |     "\n",
 62 |     "number_iter = 1000\n",
 63 |     "\n",
 64 |     "psd1D_total = np.zeros([number_iter, N])\n",
 65 |     "label_total = np.zeros([number_iter])\n",
 66 |     "psd1D_org_mean = np.zeros(N)\n",
 67 |     "psd1D_org_std = np.zeros(N)\n",
 68 |     "\n",
 69 |     "\n",
 70 |     "cont = 0\n",
 71 |     "\n",
 72 |     "#fake data\n",
 73 |     "rootdir = 'dataset_celebA/'\n",
 74 |     "\n",
 75 |     "for filename in glob.glob(rootdir+\"*.jpg\"):\n",
 76 |     "    img = cv2.imread(filename,0)\n",
 77 |     "    \n",
 78 |     "    f = np.fft.fft2(img)\n",
 79 |     "    fshift = np.fft.fftshift(f)\n",
 80 |     "    fshift += epsilon\n",
 81 |     "    \n",
 82 |     "    magnitude_spectrum = 20*np.log(np.abs(fshift))\n",
 83 |     "    psd1D = radialProfile.azimuthalAverage(magnitude_spectrum)\n",
 84 |     "\n",
 85 |     "    # Calculate the azimuthally averaged 1D power spectrum\n",
 86 |     "    points = np.linspace(0,N,num=psd1D.size) # coordinates of a\n",
 87 |     "    xi = np.linspace(0,N,num=N) # coordinates for interpolation\n",
 88 |     "\n",
 89 |     "    interpolated = griddata(points,psd1D,xi,method='cubic')\n",
 90 |     "    interpolated /= interpolated[0]\n",
 91 |     "\n",
 92 |     "    psd1D_total[cont,:] = interpolated             \n",
 93 |     "    label_total[cont] = 1\n",
 94 |     "    cont+=1\n",
 95 |     "\n",
 96 |     "    if cont == number_iter:\n",
 97 |     "        break\n",
 98 |     "\n",
 99 |     "for x in range(N):\n",
100 |     "    psd1D_org_mean[x] = np.mean(psd1D_total[:,x])\n",
101 |     "    psd1D_org_std[x]= np.std(psd1D_total[:,x])\n",
102 |     "    \n",
103 |     "\n",
104 |     "## real data\n",
105 |     "psd1D_total2 = np.zeros([number_iter, N])\n",
106 |     "label_total2 = np.zeros([number_iter])\n",
107 |     "psd1D_org_mean2 = np.zeros(N)\n",
108 |     "psd1D_org_std2 = np.zeros(N)\n",
109 |     "\n",
110 |     "cont = 0\n",
111 |     "rootdir2 = '/home/duralllopez/DATASETS/celebA/img_align_celeba/'\n",
112 |     "\n",
113 |     "\n",
114 |     "for filename in glob.glob(rootdir2+\"*.jpg\"):     \n",
115 |     "    img = cv2.imread(filename,0)\n",
116 |     "\n",
117 |     "    f = np.fft.fft2(img)\n",
118 |     "    fshift = np.fft.fftshift(f)\n",
119 |     "    fshift += epsilon\n",
120 |     "\n",
121 |     "    magnitude_spectrum = 20*np.log(np.abs(fshift))\n",
122 |     "\n",
123 |     "    # Calculate the azimuthally averaged 1D power spectrum\n",
124 |     "    psd1D = radialProfile.azimuthalAverage(magnitude_spectrum)\n",
125 |     "\n",
126 |     "    points = np.linspace(0,N,num=psd1D.size) # coordinates of a\n",
127 |     "    xi = np.linspace(0,N,num=N) # coordinates for interpolation\n",
128 |     "\n",
129 |     "    interpolated = griddata(points,psd1D,xi,method='cubic')\n",
130 |     "\n",
131 |     "    interpolated /= interpolated[0]\n",
132 |     "\n",
133 |     "    psd1D_total2[cont,:] = interpolated             \n",
134 |     "    label_total2[cont] = 0\n",
135 |     "    cont+=1\n",
136 |     "    \n",
137 |     "    if cont == number_iter:\n",
138 |     "        break\n",
139 |     "\n",
140 |     "for x in range(N):\n",
141 |     "    psd1D_org_mean2[x] = np.mean(psd1D_total2[:,x])\n",
142 |     "    psd1D_org_std2[x]= np.std(psd1D_total2[:,x])\n",
143 |     "    \n",
144 |     "\n",
145 |     "y.append(psd1D_org_mean)\n",
146 |     "y.append(psd1D_org_mean2)\n",
147 |     "error.append(psd1D_org_std)\n",
148 |     "error.append(psd1D_org_std2)\n",
149 |     "\n",
150 |     "psd1D_total_final = np.concatenate((psd1D_total,psd1D_total2), axis=0)\n",
151 |     "label_total_final = np.concatenate((label_total,label_total2), axis=0)\n",
152 |     "\n",
153 |     "data[\"data\"] = psd1D_total_final\n",
154 |     "data[\"label\"] = label_total_final\n",
155 |     "\n",
156 |     "output = open('celeba_low_1000.pkl', 'wb')\n",
157 |     "pickle.dump(data, output)\n",
158 |     "output.close()\n",
159 |     "\n",
160 |     "print(\"DATA Saved\") "
161 |    ]
162 |   },
163 |   {
164 |    "cell_type": "markdown",
165 |    "metadata": {},
166 |    "source": [
167 |     "### 2. Loading Features"
168 |    ]
169 |   },
170 |   {
171 |    "cell_type": "markdown",
172 |    "metadata": {},
173 |    "source": [
174 |     "Now, we load the features. Either the pre-computed ones or the features that you have created."
175 |    ]
176 |   },
177 |   {
178 |    "cell_type": "code",
179 |    "execution_count": 3,
180 |    "metadata": {},
181 |    "outputs": [],
182 |    "source": [
183 |     "import numpy as np\n",
184 |     "import matplotlib.pyplot as plt\n",
185 |     "import pickle\n",
186 |     "\n",
187 |     "# load feature file\n",
188 |     "pkl_file = open('celeba_low_1000.pkl', 'rb')\n",
189 |     "data = pickle.load(pkl_file)\n",
190 |     "pkl_file.close()\n",
191 |     "X = data[\"data\"]\n",
192 |     "y = data[\"label\"]"
193 |    ]
194 |   },
195 |   {
196 |    "cell_type": "markdown",
197 |    "metadata": {},
198 |    "source": [
199 |     "We look at the label distribution, to be sure that we have a balanced dataset."
200 |    ]
201 |   },
202 |   {
203 |    "cell_type": "code",
204 |    "execution_count": 4,
205 |    "metadata": {},
206 |    "outputs": [
207 |     {
208 |      "data": {
209 |       "text/plain": [
210 |        "[<matplotlib.lines.Line2D at 0x7f8cb9cf9d30>]"
211 |       ]
212 |      },
213 |      "execution_count": 4,
214 |      "metadata": {},
215 |      "output_type": "execute_result"
216 |     },
217 |     {
218 |      "data": {
219 |       "image/png": "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\n",
220 |       "text/plain": [
221 |        "<Figure size 432x288 with 1 Axes>"
222 |       ]
223 |      },
224 |      "metadata": {
225 |       "needs_background": "light"
226 |      },
227 |      "output_type": "display_data"
228 |     }
229 |    ],
230 |    "source": [
231 |     "plt.plot(y)"
232 |    ]
233 |   },
234 |   {
235 |    "cell_type": "markdown",
236 |    "metadata": {},
237 |    "source": [
238 |     "### 3. Check Spectrum"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "markdown",
243 |    "metadata": {},
244 |    "source": [
245 |     "We have a look to the spectrum"
246 |    ]
247 |   },
248 |   {
249 |    "cell_type": "code",
250 |    "execution_count": 5,
251 |    "metadata": {},
252 |    "outputs": [
253 |     {
254 |      "data": {
255 |       "text/plain": [
256 |        "Text(0, 0.5, 'Power Spectrum')"
257 |       ]
258 |      },
259 |      "execution_count": 5,
260 |      "metadata": {},
261 |      "output_type": "execute_result"
262 |     },
263 |     {
264 |      "data": {
265 |       "image/png": "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\n",
266 |       "text/plain": [
267 |        "<Figure size 1080x648 with 1 Axes>"
268 |       ]
269 |      },
270 |      "metadata": {
271 |       "needs_background": "light"
272 |      },
273 |      "output_type": "display_data"
274 |     }
275 |    ],
276 |    "source": [
277 |     "num = int(X.shape[0]/2)\n",
278 |     "num_feat = X.shape[1]\n",
279 |     "\n",
280 |     "psd1D_org_0 = np.zeros((num,num_feat))\n",
281 |     "psd1D_org_1 = np.zeros((num,num_feat))\n",
282 |     "psd1D_org_0_mean = np.zeros(num_feat)\n",
283 |     "psd1D_org_0_std = np.zeros(num_feat)\n",
284 |     "psd1D_org_1_mean = np.zeros(num_feat)\n",
285 |     "psd1D_org_1_std = np.zeros(num_feat)\n",
286 |     "\n",
287 |     "cont_0=0\n",
288 |     "cont_1=0\n",
289 |     "\n",
290 |     "# We separate real and fake using the label\n",
291 |     "for x in range(X.shape[0]):\n",
292 |     "    if y[x]==0:\n",
293 |     "        psd1D_org_0[cont_0,:] = X[x,:]\n",
294 |     "        cont_0+=1\n",
295 |     "    elif y[x]==1:\n",
296 |     "        psd1D_org_1[cont_1,:] = X[x,:]\n",
297 |     "        cont_1+=1\n",
298 |     "\n",
299 |     "# We compute statistcis\n",
300 |     "for x in range(num_feat):\n",
301 |     "    psd1D_org_0_mean[x] = np.mean(psd1D_org_0[:,x])\n",
302 |     "    psd1D_org_0_std[x]= np.std(psd1D_org_0[:,x])\n",
303 |     "    psd1D_org_1_mean[x] = np.mean(psd1D_org_1[:,x])\n",
304 |     "    psd1D_org_1_std[x]= np.std(psd1D_org_1[:,x])\n",
305 |     "    \n",
306 |     "# Plot\n",
307 |     "x = np.arange(0, num_feat, 1)\n",
308 |     "fig, ax = plt.subplots(figsize=(15, 9))\n",
309 |     "ax.plot(x, psd1D_org_0_mean, alpha=0.5, color='red', label='Fake', linewidth =2.0)\n",
310 |     "ax.fill_between(x, psd1D_org_0_mean - psd1D_org_0_std, psd1D_org_0_mean + psd1D_org_0_std, color='red', alpha=0.2)\n",
311 |     "ax.plot(x, psd1D_org_1_mean, alpha=0.5, color='blue', label='Real', linewidth =2.0)\n",
312 |     "ax.fill_between(x, psd1D_org_1_mean - psd1D_org_1_std, psd1D_org_1_mean + psd1D_org_1_std, color='blue', alpha=0.2)\n",
313 |     "\n",
314 |     "plt.tick_params(axis='x', labelsize=20)\n",
315 |     "plt.tick_params(axis='y', labelsize=20)\n",
316 |     "ax.legend(loc='best', prop={'size': 20})\n",
317 |     "plt.xlabel(\"Spatial Frequency\", fontsize=20)\n",
318 |     "plt.ylabel(\"Power Spectrum\", fontsize=20)\n",
319 |     "#plt.savefig('1000_celeba.png', bbox_inches='tight')"
320 |    ]
321 |   },
322 |   {
323 |    "cell_type": "markdown",
324 |    "metadata": {},
325 |    "source": [
326 |     "### 4. Classification"
327 |    ]
328 |   },
329 |   {
330 |    "cell_type": "markdown",
331 |    "metadata": {},
332 |    "source": [
333 |     "Now we classify using the features."
334 |    ]
335 |   },
336 |   {
337 |    "cell_type": "code",
338 |    "execution_count": 7,
339 |    "metadata": {},
340 |    "outputs": [
341 |     {
342 |      "name": "stderr",
343 |      "output_type": "stream",
344 |      "text": [
345 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
346 |       "  \"avoid this warning.\", FutureWarning)\n",
347 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
348 |       "  \"avoid this warning.\", FutureWarning)\n",
349 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
350 |       "  \"avoid this warning.\", FutureWarning)\n",
351 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
352 |       "  \"avoid this warning.\", FutureWarning)\n"
353 |      ]
354 |     },
355 |     {
356 |      "name": "stdout",
357 |      "output_type": "stream",
358 |      "text": [
359 |       "Average SVM: 0.9995\n",
360 |       "Average SVM_r: 1.0\n",
361 |       "Average SVM_p: 0.9855\n",
362 |       "Average LR: 0.9975000000000002\n"
363 |      ]
364 |     },
365 |     {
366 |      "name": "stderr",
367 |      "output_type": "stream",
368 |      "text": [
369 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
370 |       "  \"avoid this warning.\", FutureWarning)\n"
371 |      ]
372 |     }
373 |    ],
374 |    "source": [
375 |     "import numpy as np\n",
376 |     "import matplotlib.pyplot as plt\n",
377 |     "import pickle\n",
378 |     "\n",
379 |     "num = 10\n",
380 |     "LR = 0\n",
381 |     "SVM = 0\n",
382 |     "SVM_r = 0\n",
383 |     "SVM_p = 0\n",
384 |     "\n",
385 |     "\n",
386 |     "for z in range(num):\n",
387 |     "    # read python dict back from the file\n",
388 |     "    pkl_file = open('celeba_low_1000.pkl', 'rb')\n",
389 |     "    \n",
390 |     "    data = pickle.load(pkl_file)\n",
391 |     "\n",
392 |     "    pkl_file.close()\n",
393 |     "    X = data[\"data\"]\n",
394 |     "    y = data[\"label\"]\n",
395 |     "\n",
396 |     "\n",
397 |     "    try:\n",
398 |     "\n",
399 |     "        from sklearn.model_selection import train_test_split\n",
400 |     "        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)\n",
401 |     "\n",
402 |     "        from sklearn.svm import SVC\n",
403 |     "        svclassifier = SVC(kernel='linear')\n",
404 |     "        svclassifier.fit(X_train, y_train)\n",
405 |     "        #print('Accuracy on test set: {:.3f}'.format(svclassifier.score(X_test, y_test)))\n",
406 |     "               \n",
407 |     "        from sklearn.svm import SVC\n",
408 |     "        svclassifier_r = SVC(C=6.37, kernel='rbf', gamma=0.86)\n",
409 |     "        svclassifier_r.fit(X_train, y_train)\n",
410 |     "        #print('Accuracy on test set: {:.3f}'.format(svclassifier_r.score(X_test, y_test)))\n",
411 |     "                \n",
412 |     "        from sklearn.svm import SVC\n",
413 |     "        svclassifier_p = SVC(kernel='poly')\n",
414 |     "        svclassifier_p.fit(X_train, y_train)\n",
415 |     "        #print('Accuracy on test set: {:.3f}'.format(svclassifier_p.score(X_test, y_test)))\n",
416 |     "        \n",
417 |     "        from sklearn.linear_model import LogisticRegression\n",
418 |     "        logreg = LogisticRegression(solver='liblinear', max_iter=1000)\n",
419 |     "        logreg.fit(X_train, y_train)\n",
420 |     "        #print('Accuracy on test set: {:.3f}'.format(logreg.score(X_test, y_test)))\n",
421 |     "\n",
422 |     "        \n",
423 |     "        SVM+=svclassifier.score(X_test, y_test)\n",
424 |     "        SVM_r+=svclassifier_r.score(X_test, y_test)\n",
425 |     "        SVM_p+=svclassifier_p.score(X_test, y_test)\n",
426 |     "        LR+=logreg.score(X_test, y_test)\n",
427 |     "\n",
428 |     "        \n",
429 |     "    except:\n",
430 |     "        num-=1\n",
431 |     "        print(num)\n",
432 |     "    \n",
433 |     "print(\"Average SVM: \"+str(SVM/num))\n",
434 |     "print(\"Average SVM_r: \"+str(SVM_r/num))\n",
435 |     "print(\"Average SVM_p: \"+str(SVM_p/num))\n",
436 |     "print(\"Average LR: \"+str(LR/num))"
437 |    ]
438 |   },
439 |   {
440 |    "cell_type": "code",
441 |    "execution_count": null,
442 |    "metadata": {},
443 |    "outputs": [],
444 |    "source": []
445 |   }
446 |  ],
447 |  "metadata": {
448 |   "kernelspec": {
449 |    "display_name": "Python 3",
450 |    "language": "python",
451 |    "name": "python3"
452 |   },
453 |   "language_info": {
454 |    "codemirror_mode": {
455 |     "name": "ipython",
456 |     "version": 3
457 |    },
458 |    "file_extension": ".py",
459 |    "mimetype": "text/x-python",
460 |    "name": "python",
461 |    "nbconvert_exporter": "python",
462 |    "pygments_lexer": "ipython3",
463 |    "version": "3.6.7"
464 |   }
465 |  },
466 |  "nbformat": 4,
467 |  "nbformat_minor": 4
468 | }
469 | 


--------------------------------------------------------------------------------
/Experiments_CelebA/README.md:
--------------------------------------------------------------------------------
1 | # Experiments CelebA
2 | <img align="center" src="/imgs/1000_celeba.png" width="800"/>
3 | 
4 | 
5 | 


--------------------------------------------------------------------------------
/Experiments_CelebA/celeba_low_1000.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_CelebA/celeba_low_1000.pkl


--------------------------------------------------------------------------------
/Experiments_CelebA/dataset_celebA.7z:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_CelebA/dataset_celebA.7z


--------------------------------------------------------------------------------
/Experiments_DeepFakeDetection/.gitkeep:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_DeepFakeDetection/.gitkeep


--------------------------------------------------------------------------------
/Experiments_DeepFakeDetection/README.md:
--------------------------------------------------------------------------------
1 | # Experiments DeepFakeDetection
2 | You can downlaod pre-processing FaceForensics++ images [link](https://bit.ly/2wkPZYv) 
3 | <img align="center" src="/imgs/1000_deep.png" width="800"/>
4 | 
5 | 
6 | 


--------------------------------------------------------------------------------
/Experiments_DeepFakeDetection/radialProfile.py:
--------------------------------------------------------------------------------
 1 | # from https://www.astrobetter.com/blog/2010/03/03/fourier-transforms-of-images-in-python/
 2 | import numpy as np
 3 | 
 4 | def azimuthalAverage(image, center=None):
 5 |     """
 6 |     Calculate the azimuthally averaged radial profile.
 7 | 
 8 |     image - The 2D image
 9 |     center - The [x,y] pixel coordinates used as the center. The default is 
10 |              None, which then uses the center of the image (including 
11 |              fracitonal pixels).
12 |     
13 |     """
14 |     # Calculate the indices from the image
15 |     y, x = np.indices(image.shape)
16 | 
17 |     if not center:
18 |         center = np.array([(x.max()-x.min())/2.0, (y.max()-y.min())/2.0])
19 | 
20 |     r = np.hypot(x - center[0], y - center[1])
21 | 
22 |     # Get sorted radii
23 |     ind = np.argsort(r.flat)
24 |     r_sorted = r.flat[ind]
25 |     i_sorted = image.flat[ind]
26 | 
27 |     # Get the integer part of the radii (bin size = 1)
28 |     r_int = r_sorted.astype(int)
29 | 
30 |     # Find all pixels that fall within each radial bin.
31 |     deltar = r_int[1:] - r_int[:-1]  # Assumes all radii represented
32 |     rind = np.where(deltar)[0]       # location of changed radius
33 |     nr = rind[1:] - rind[:-1]        # number of radius bin
34 |     
35 |     # Cumulative sum to figure out sums for each radius bin
36 |     csim = np.cumsum(i_sorted, dtype=float)
37 |     tbin = csim[rind[1:]] - csim[rind[:-1]]
38 | 
39 |     radial_prof = tbin / nr
40 | 
41 |     return radial_prof
42 | 


--------------------------------------------------------------------------------
/Experiments_DeepFakeDetection/test_1000.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_DeepFakeDetection/test_1000.pkl


--------------------------------------------------------------------------------
/Experiments_DeepFakeDetection/train_3200.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_DeepFakeDetection/train_3200.pkl


--------------------------------------------------------------------------------
/Experiments_Faces-HQ/.gitkeep:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_Faces-HQ/.gitkeep


--------------------------------------------------------------------------------
/Experiments_Faces-HQ/Faces-HQ.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# Faces-HQ "
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "markdown",
 12 |    "metadata": {},
 13 |    "source": [
 14 |     "In this notebook we show the results for Faces-HQ. You can create from scratch the features or use the pre-computed ones."
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "markdown",
 19 |    "metadata": {},
 20 |    "source": [
 21 |     "### 1. Create feature"
 22 |    ]
 23 |   },
 24 |   {
 25 |    "cell_type": "markdown",
 26 |    "metadata": {},
 27 |    "source": [
 28 |     "If you want to create the features, first of all download the data from [link](https://cutt.ly/6enDLYG). Be sure to save the folder together with this notebook. \n",
 29 |     "\n",
 30 |     "Otherwise, just jump to section 2."
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 1,
 36 |    "metadata": {},
 37 |    "outputs": [
 38 |     {
 39 |      "name": "stdout",
 40 |      "output_type": "stream",
 41 |      "text": [
 42 |       "thispersondoesntexists\n",
 43 |       "100KFake\n",
 44 |       "Flickr-Faces-HQ2_\n",
 45 |       "celebA-HQ_10K\n",
 46 |       "DATA Saved\n"
 47 |      ]
 48 |     }
 49 |    ],
 50 |    "source": [
 51 |     "import cv2\n",
 52 |     "import numpy as np\n",
 53 |     "import os\n",
 54 |     "import radialProfile\n",
 55 |     "import glob\n",
 56 |     "from matplotlib import pyplot as plt\n",
 57 |     "import pickle\n",
 58 |     "\n",
 59 |     "\n",
 60 |     "path = ['thispersondoesntexists', '100KFake','Flickr-Faces-HQ2_', 'celebA-HQ_10K']\n",
 61 |     "labels = [1,1,0,0]\n",
 62 |     "format_file = ['jpg','jpg','jpg', 'jpg']\n",
 63 |     "epsilon = 1e-8\n",
 64 |     "data = {}\n",
 65 |     "#number of samples from each dataset\n",
 66 |     "stop = 250\n",
 67 |     "number_iter = 4 * stop\n",
 68 |     "psd1D_total = np.zeros([number_iter, 722])\n",
 69 |     "label_total = np.zeros([number_iter])\n",
 70 |     "iter_ = 0\n",
 71 |     "\n",
 72 |     "for z in range(4):\n",
 73 |     "    cont = 0\n",
 74 |     "    psd1D_average_org = np.zeros(722)\n",
 75 |     "    print(path[z])\n",
 76 |     "    \n",
 77 |     "    for filename in glob.glob(path[z]+\"/*.\"+format_file[z]):  \n",
 78 |     "        img = cv2.imread(filename,0)\n",
 79 |     "        f = np.fft.fft2(img)\n",
 80 |     "        fshift = np.fft.fftshift(f)\n",
 81 |     "        fshift += epsilon\n",
 82 |     "\n",
 83 |     "        magnitude_spectrum = 20*np.log(np.abs(fshift))\n",
 84 |     "\n",
 85 |     "        # Calculate the azimuthally averaged 1D power spectrum\n",
 86 |     "        psd1D = radialProfile.azimuthalAverage(magnitude_spectrum)\n",
 87 |     "        psd1D_total[iter_,:] = psd1D\n",
 88 |     "        label_total[iter_] = labels[z]\n",
 89 |     "\n",
 90 |     "        cont+=1\n",
 91 |     "        iter_+=1\n",
 92 |     "        if cont >= stop:\n",
 93 |     "            break\n",
 94 |     "\n",
 95 |     "data[\"data\"] = psd1D_total\n",
 96 |     "data[\"label\"] = label_total\n",
 97 |     "\n",
 98 |     "output = open('dataset_freq_1000.pkl', 'wb')\n",
 99 |     "pickle.dump(data, output)\n",
100 |     "output.close()\n",
101 |     "print(\"DATA Saved\")    "
102 |    ]
103 |   },
104 |   {
105 |    "cell_type": "markdown",
106 |    "metadata": {},
107 |    "source": [
108 |     "### 2. Loading Features"
109 |    ]
110 |   },
111 |   {
112 |    "cell_type": "markdown",
113 |    "metadata": {},
114 |    "source": [
115 |     "Now, we load the features. Either the pre-computed ones or the features that you have created."
116 |    ]
117 |   },
118 |   {
119 |    "cell_type": "code",
120 |    "execution_count": 18,
121 |    "metadata": {},
122 |    "outputs": [],
123 |    "source": [
124 |     "import numpy as np\n",
125 |     "import matplotlib.pyplot as plt\n",
126 |     "import pickle\n",
127 |     "\n",
128 |     "# load feature file\n",
129 |     "pkl_file = open('dataset_freq_1000.pkl', 'rb')\n",
130 |     "data = pickle.load(pkl_file)\n",
131 |     "pkl_file.close()\n",
132 |     "X = data[\"data\"]\n",
133 |     "y = data[\"label\"]"
134 |    ]
135 |   },
136 |   {
137 |    "cell_type": "markdown",
138 |    "metadata": {},
139 |    "source": [
140 |     "We look at the label distribution, to be sure that we have a balanced dataset."
141 |    ]
142 |   },
143 |   {
144 |    "cell_type": "code",
145 |    "execution_count": 19,
146 |    "metadata": {},
147 |    "outputs": [
148 |     {
149 |      "data": {
150 |       "text/plain": [
151 |        "[<matplotlib.lines.Line2D at 0x7f58511d1e10>]"
152 |       ]
153 |      },
154 |      "execution_count": 19,
155 |      "metadata": {},
156 |      "output_type": "execute_result"
157 |     },
158 |     {
159 |      "data": {
160 |       "image/png": "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\n",
161 |       "text/plain": [
162 |        "<Figure size 432x288 with 1 Axes>"
163 |       ]
164 |      },
165 |      "metadata": {
166 |       "needs_background": "light"
167 |      },
168 |      "output_type": "display_data"
169 |     }
170 |    ],
171 |    "source": [
172 |     "plt.plot(y)"
173 |    ]
174 |   },
175 |   {
176 |    "cell_type": "markdown",
177 |    "metadata": {},
178 |    "source": [
179 |     "### 3. Check Spectrum"
180 |    ]
181 |   },
182 |   {
183 |    "cell_type": "markdown",
184 |    "metadata": {},
185 |    "source": [
186 |     "We have a look to the spectrum"
187 |    ]
188 |   },
189 |   {
190 |    "cell_type": "code",
191 |    "execution_count": 21,
192 |    "metadata": {},
193 |    "outputs": [
194 |     {
195 |      "data": {
196 |       "text/plain": [
197 |        "Text(0, 0.5, 'Power Spectrum')"
198 |       ]
199 |      },
200 |      "execution_count": 21,
201 |      "metadata": {},
202 |      "output_type": "execute_result"
203 |     },
204 |     {
205 |      "data": {
206 |       "image/png": "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\n",
207 |       "text/plain": [
208 |        "<Figure size 1080x648 with 1 Axes>"
209 |       ]
210 |      },
211 |      "metadata": {
212 |       "needs_background": "light"
213 |      },
214 |      "output_type": "display_data"
215 |     }
216 |    ],
217 |    "source": [
218 |     "num = int(X.shape[0]/2)\n",
219 |     "num_feat = X.shape[1]\n",
220 |     "\n",
221 |     "psd1D_org_0 = np.zeros((num,num_feat))\n",
222 |     "psd1D_org_1 = np.zeros((num,num_feat))\n",
223 |     "psd1D_org_0_mean = np.zeros(num_feat)\n",
224 |     "psd1D_org_0_std = np.zeros(num_feat)\n",
225 |     "psd1D_org_1_mean = np.zeros(num_feat)\n",
226 |     "psd1D_org_1_std = np.zeros(num_feat)\n",
227 |     "\n",
228 |     "cont_0=0\n",
229 |     "cont_1=0\n",
230 |     "\n",
231 |     "# We separate real and fake using the label\n",
232 |     "for x in range(X.shape[0]):\n",
233 |     "    if y[x]==0:\n",
234 |     "        psd1D_org_0[cont_0,:] = X[x,:]\n",
235 |     "        cont_0+=1\n",
236 |     "    elif y[x]==1:\n",
237 |     "        psd1D_org_1[cont_1,:] = X[x,:]\n",
238 |     "        cont_1+=1\n",
239 |     "\n",
240 |     "# We compute statistcis\n",
241 |     "for x in range(num_feat):\n",
242 |     "    psd1D_org_0_mean[x] = np.mean(psd1D_org_0[:,x])\n",
243 |     "    psd1D_org_0_std[x]= np.std(psd1D_org_0[:,x])\n",
244 |     "    psd1D_org_1_mean[x] = np.mean(psd1D_org_1[:,x])\n",
245 |     "    psd1D_org_1_std[x]= np.std(psd1D_org_1[:,x])\n",
246 |     "    \n",
247 |     "# Plot\n",
248 |     "x = np.arange(0, num_feat, 1)\n",
249 |     "fig, ax = plt.subplots(figsize=(15, 9))\n",
250 |     "ax.plot(x, psd1D_org_0_mean, alpha=0.5, color='red', label='Real', linewidth =2.0)\n",
251 |     "ax.fill_between(x, psd1D_org_0_mean - psd1D_org_0_std, psd1D_org_0_mean + psd1D_org_0_std, color='red', alpha=0.2)\n",
252 |     "ax.plot(x, psd1D_org_1_mean, alpha=0.5, color='blue', label='Fake', linewidth =2.0)\n",
253 |     "ax.fill_between(x, psd1D_org_1_mean - psd1D_org_1_std, psd1D_org_1_mean + psd1D_org_1_std, color='blue', alpha=0.2)\n",
254 |     "ax.legend()\n",
255 |     "plt.tick_params(axis='x', labelsize=20)\n",
256 |     "plt.tick_params(axis='y', labelsize=20)\n",
257 |     "ax.legend(loc='best', prop={'size': 20})\n",
258 |     "plt.xlabel(\"Spatial Frequency\", fontsize=20)\n",
259 |     "plt.ylabel(\"Power Spectrum\", fontsize=20)"
260 |    ]
261 |   },
262 |   {
263 |    "cell_type": "markdown",
264 |    "metadata": {},
265 |    "source": [
266 |     "### 4. Classification"
267 |    ]
268 |   },
269 |   {
270 |    "cell_type": "markdown",
271 |    "metadata": {},
272 |    "source": [
273 |     "Now we classify using the features."
274 |    ]
275 |   },
276 |   {
277 |    "cell_type": "code",
278 |    "execution_count": 22,
279 |    "metadata": {},
280 |    "outputs": [
281 |     {
282 |      "name": "stderr",
283 |      "output_type": "stream",
284 |      "text": [
285 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
286 |       "  \"avoid this warning.\", FutureWarning)\n",
287 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
288 |       "  \"avoid this warning.\", FutureWarning)\n",
289 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
290 |       "  \"avoid this warning.\", FutureWarning)\n",
291 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
292 |       "  \"avoid this warning.\", FutureWarning)\n",
293 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
294 |       "  \"avoid this warning.\", FutureWarning)\n",
295 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
296 |       "  \"avoid this warning.\", FutureWarning)\n",
297 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
298 |       "  \"avoid this warning.\", FutureWarning)\n",
299 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
300 |       "  \"avoid this warning.\", FutureWarning)\n",
301 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
302 |       "  \"avoid this warning.\", FutureWarning)\n"
303 |      ]
304 |     },
305 |     {
306 |      "name": "stdout",
307 |      "output_type": "stream",
308 |      "text": [
309 |       "Average SVM: 1.0\n",
310 |       "Average LR: 1.0\n"
311 |      ]
312 |     },
313 |     {
314 |      "name": "stderr",
315 |      "output_type": "stream",
316 |      "text": [
317 |       "/opt/anaconda3/lib/python3.6/site-packages/sklearn/svm/base.py:193: FutureWarning: The default value of gamma will change from 'auto' to 'scale' in version 0.22 to account better for unscaled features. Set gamma explicitly to 'auto' or 'scale' to avoid this warning.\n",
318 |       "  \"avoid this warning.\", FutureWarning)\n"
319 |      ]
320 |     }
321 |    ],
322 |    "source": [
323 |     "import numpy as np\n",
324 |     "import matplotlib.pyplot as plt\n",
325 |     "import pickle\n",
326 |     "\n",
327 |     "num = 10\n",
328 |     "LR = 0\n",
329 |     "SVM = 0\n",
330 |     "\n",
331 |     "\n",
332 |     "for z in range(num):\n",
333 |     "    # read python dict back from the file\n",
334 |     "    pkl_file = open('dataset_freq_1000.pkl', 'rb')\n",
335 |     "    \n",
336 |     "    data = pickle.load(pkl_file)\n",
337 |     "\n",
338 |     "    pkl_file.close()\n",
339 |     "    X = data[\"data\"]\n",
340 |     "    y = data[\"label\"]\n",
341 |     "\n",
342 |     "    try:\n",
343 |     "\n",
344 |     "        from sklearn.model_selection import train_test_split\n",
345 |     "        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)\n",
346 |     "\n",
347 |     "        from sklearn.svm import SVC\n",
348 |     "        svclassifier = SVC(kernel='linear')\n",
349 |     "        svclassifier.fit(X_train, y_train)\n",
350 |     "        #print('Accuracy on test set: {:.3f}'.format(svclassifier.score(X_test, y_test)))\n",
351 |     "        \n",
352 |     "        from sklearn.linear_model import LogisticRegression\n",
353 |     "        logreg = LogisticRegression(solver='liblinear', max_iter=1000)\n",
354 |     "        logreg.fit(X_train, y_train)\n",
355 |     "        #print('Accuracy on test set: {:.3f}'.format(logreg.score(X_test, y_test)))\n",
356 |     "\n",
357 |     "        \n",
358 |     "        SVM+=svclassifier.score(X_test, y_test)\n",
359 |     "        LR+=logreg.score(X_test, y_test)\n",
360 |     "\n",
361 |     " \n",
362 |     "    except:\n",
363 |     "        num-=1\n",
364 |     "        print(num)\n",
365 |     "    \n",
366 |     "print(\"Average SVM: \"+str(SVM/num))\n",
367 |     "print(\"Average LR: \"+str(LR/num))"
368 |    ]
369 |   },
370 |   {
371 |    "cell_type": "code",
372 |    "execution_count": null,
373 |    "metadata": {},
374 |    "outputs": [],
375 |    "source": []
376 |   }
377 |  ],
378 |  "metadata": {
379 |   "kernelspec": {
380 |    "display_name": "Python 3",
381 |    "language": "python",
382 |    "name": "python3"
383 |   },
384 |   "language_info": {
385 |    "codemirror_mode": {
386 |     "name": "ipython",
387 |     "version": 3
388 |    },
389 |    "file_extension": ".py",
390 |    "mimetype": "text/x-python",
391 |    "name": "python",
392 |    "nbconvert_exporter": "python",
393 |    "pygments_lexer": "ipython3",
394 |    "version": "3.6.7"
395 |   }
396 |  },
397 |  "nbformat": 4,
398 |  "nbformat_minor": 4
399 | }
400 | 


--------------------------------------------------------------------------------
/Experiments_Faces-HQ/README.md:
--------------------------------------------------------------------------------
1 | # Experiments Face-HQ
2 | <img align="center" src="/imgs/1000_hq.png" width="800"/>


--------------------------------------------------------------------------------
/Experiments_Faces-HQ/dataset_freq_1000.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/Experiments_Faces-HQ/dataset_freq_1000.pkl


--------------------------------------------------------------------------------
/Experiments_Faces-HQ/radialProfile.py:
--------------------------------------------------------------------------------
 1 | # from https://www.astrobetter.com/blog/2010/03/03/fourier-transforms-of-images-in-python/
 2 | import numpy as np
 3 | 
 4 | def azimuthalAverage(image, center=None):
 5 |     """
 6 |     Calculate the azimuthally averaged radial profile.
 7 | 
 8 |     image - The 2D image
 9 |     center - The [x,y] pixel coordinates used as the center. The default is 
10 |              None, which then uses the center of the image (including 
11 |              fracitonal pixels).
12 |     
13 |     """
14 |     # Calculate the indices from the image
15 |     y, x = np.indices(image.shape)
16 | 
17 |     if not center:
18 |         center = np.array([(x.max()-x.min())/2.0, (y.max()-y.min())/2.0])
19 | 
20 |     r = np.hypot(x - center[0], y - center[1])
21 | 
22 |     # Get sorted radii
23 |     ind = np.argsort(r.flat)
24 |     r_sorted = r.flat[ind]
25 |     i_sorted = image.flat[ind]
26 | 
27 |     # Get the integer part of the radii (bin size = 1)
28 |     r_int = r_sorted.astype(int)
29 | 
30 |     # Find all pixels that fall within each radial bin.
31 |     deltar = r_int[1:] - r_int[:-1]  # Assumes all radii represented
32 |     rind = np.where(deltar)[0]       # location of changed radius
33 |     nr = rind[1:] - rind[:-1]        # number of radius bin
34 |     
35 |     # Cumulative sum to figure out sums for each radius bin
36 |     csim = np.cumsum(i_sorted, dtype=float)
37 |     tbin = csim[rind[1:]] - csim[rind[:-1]]
38 | 
39 |     radial_prof = tbin / nr
40 | 
41 |     return radial_prof
42 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
  1 | # [Unmasking DeepFake with simple Features](https://arxiv.org/abs/1911.00686)
  2 | 
  3 | This repository provides the official Python implementation of Unmasking DeepFake with simple Features (Paper: [https://arxiv.org/abs/1911.00686](https://arxiv.org/abs/1911.00686)).
  4 | 
  5 | <img align="center" src="imgs/pipeline.png" width="1000"/>
  6 | 
  7 | Overview of the pipeline used in our approach. It contains two main blocks, a pre-processing where the input istransformed to a more convenient domain and a 
  8 | training block, where a classifier uses the new transformed features to determine whether the face is real or not. Notice that input images are grey-scaled 
  9 | before DFT.
 10 | 
 11 | ## Dependencies
 12 | Tested on Python 3.6.x.
 13 | * [NumPy](http://www.numpy.org/) (1.16.2)
 14 | * [Opencv](https://opencv.org/opencv-4-0/) (4.0.0)
 15 | * [Matplotlib](https://matplotlib.org/) (3.1.1)
 16 | 
 17 | 
 18 | 
 19 | ## Detection Faces-HQ 
 20 | To the best of our knowledge, no public dataset gathers images containing both artificially and real faces, therefore, we have created our own called Faces-HQ.
 21 | In order to have a sufficient variety of faces, we have chosen to download and label, images available from [CelebA-HQ dataset](https://arxiv.org/abs/1710.10196),
 22 | [Flickr-Faces-HQ dataset](https://arxiv.org/abs/1812.04948), [100K Facesproject](https://generated.photos/) and [www.thispersondoesnotexist.com](www.thispersondoesnotexist.com). 
 23 | In total, we have collected 40K high quality images being half of them real and the other half fake faces, achieving in this manner a balanced dataset.
 24 | 
 25 | Click [here](/Experiments_Faces-HQ) to go the experiments on Faces-HQ.
 26 | 
 27 | ### Results
 28 | 
 29 | 
 30 | <b>Faces-HQ dataset.</b>
 31 | Test accuracy using SVM, logistic regression and k-means classifier under different data settings.
 32 | 
 33 | <p align='center'>  
 34 |     <img align="center" src="imgs/faces_results.PNG" width="300"/>
 35 | </p>
 36 | 
 37 | 
 38 | 
 39 | ### Detection CelebA
 40 |  [CelebA](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html) CelebFaces Attributes Dataset (CelebA) is a large-scale face attributes dataset with more than 
 41 | 200K celebrity images, each with 40 attribute annotations. The images in this dataset cover large pose variations and background clutter. CelebA has large 
 42 | diversities, large quantities, and rich annotations
 43 | 
 44 | Click [here](/Experiments_CelebA) to go the experiments on CelebA.
 45 | 
 46 | ### Results
 47 | <p align='center'>  
 48 |     <img align="center" src="imgs/celeba_results.PNG" width="300"/>
 49 | </p>
 50 | 
 51 | 
 52 | 
 53 | ## Detection DeepFakeDetection (FaceForensics++)
 54 |  [FaceForensics++](https://github.com/ondyari/FaceForensics) is a forensics dataset consisting of video sequences that have been modified with
 55 | different automated face manipulation methods. Additionally,it is hosting DeepFakeDetection Dataset. In particular, this dataset contains 363 original
 56 | sequences from 28 paid actors in 16 different scenes as well as over 3000 manipulated videos using DeepFakes and their corresponding binary masks.
 57 | All videos contain a trackable mostly frontal face without occlusions which enables automated tampering methods to generate realistic forgeries.
 58 | 
 59 | Click [here](/Experiments_DeepFakeDetection) to go the experiments on DeepFakeDetection.
 60 | 
 61 | ### Results
 62 | 
 63 | <b>DeepFakeDetection dataset.</b>
 64 | <p><i>Results based on frames.<i></p>
 65 | <p>Test accuracy using SVM and logistic regression classifier under different data settings.</p>
 66 | <p align='center'>  
 67 |     <img align="center" src="imgs/deep_results.PNG" width="300"/>
 68 | </p>
 69 | 
 70 | 
 71 | <p><i>Results based on videos. (We apply a simple majority vote over the single frame classifications).<i></p>
 72 | <p>Test accuracy using SVM and logistic regression classifier.</p>
 73 | <p align='center'>  
 74 |     <img align="center" src="imgs/results3.png" width="200"/>
 75 | </p>
 76 | 
 77 | 
 78 | 
 79 | ## Datasets Faces-HQ 
 80 | 
 81 | This repo uses and combines several datasets to form Faces-HQ:
 82 | 
 83 | <img align="center" src="imgs/dataset.png" width="500"/>
 84 | 
 85 | >We take 10K samples from [CelebA-HQ dataset](https://arxiv.org/abs/1710.10196). 
 86 | 
 87 | >We take 10K samples from [Flickr-Faces-HQ dataset](https://arxiv.org/abs/1812.04948)
 88 | and we convert to JPEG format.
 89 |  
 90 | >We take 10K samples from [www.thispersondoesnotexist.com](www.thispersondoesnotexist.com) uisng this 
 91 | [script](https://github.com/rayheffer/tpdne/blob/master/tpdne.sh)
 92 | 
 93 | >We take 10K samples from [100K Facesproject](https://generated.photos/).
 94 | 
 95 | ### Download full (19GB) Faces-HQ data set: [https://cutt.ly/6enDLYG](https://cutt.ly/6enDLYG)
 96 | 
 97 | 
 98 | ###  Citation
 99 | If this work is useful for your research, please cite our [paper](https://arxiv.org/abs/1911.00686):
100 | ```
101 | @misc{durall2019unmasking,
102 |     title={Unmasking DeepFakes with simple Features},
103 |     author={Ricard Durall and Margret Keuper and Franz-Josef Pfreundt and Janis Keuper},
104 |     year={2019},
105 |     eprint={1911.00686},
106 |     archivePrefix={arXiv},
107 |     primaryClass={cs.LG}
108 | }
109 | ```
110 | 
111 | ## Some notes on data pre-processing
112 | Some users have difficulties to get the deteection working on new data sets. Here are some remarks:
113 | * For complex scenes, you need to run a feace detection first! Our approach will not work if the face/fake is not the dominant part of the input. Try to capture the inner parts of the faces without a lot of background... 
114 | * Any re-sampling/re-scaling of the input images might distort the frequency spectrum: Do NOT resize the images, resize the spectra afterwards! Also: some prominent face detectors do resizing, don't use them if you can't turn it off.
115 | * Use square input images (non square image might distort the radial sampling)
116 | * Plot the spectra of your input data to check if they show the charaecteristic propoerties
117 | * Our approach might not work on videos/images that have been compressed to a large extend (impacts the spectrum). 
118 | 
119 | # Follow-up work (CVPR Paper)
120 | Following this pre-print, we have a CVPR 2020 paper, looking into the theory of spectral distortions by GANs and a way to fix this. 
121 | 
122 | * Pre-Print: [Watch your Up-Convolution: CNN Based Generative Deep Neural Networks are Failing to Reproduce Spectral Distributions](https://arxiv.org/abs/2003.01826)
123 | * [GitHub Repo](https://github.com/cc-hpc-itwm/UpConv)
124 | 


--------------------------------------------------------------------------------
/imgs/.gitkeep:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/.gitkeep


--------------------------------------------------------------------------------
/imgs/1000_celeba.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/1000_celeba.png


--------------------------------------------------------------------------------
/imgs/1000_deep.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/1000_deep.png


--------------------------------------------------------------------------------
/imgs/1000_hq.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/1000_hq.png


--------------------------------------------------------------------------------
/imgs/celeba_results.PNG:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/celeba_results.PNG


--------------------------------------------------------------------------------
/imgs/dataset.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/dataset.png


--------------------------------------------------------------------------------
/imgs/deep_results.PNG:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/deep_results.PNG


--------------------------------------------------------------------------------
/imgs/faces_results.PNG:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/faces_results.PNG


--------------------------------------------------------------------------------
/imgs/pipeline.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/pipeline.png


--------------------------------------------------------------------------------
/imgs/results3.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/cc-hpc-itwm/DeepFakeDetection/aaff429e0a90c09176dbcb7a754bbfc34b938741/imgs/results3.png


--------------------------------------------------------------------------------