Deepfakes Datasets
Published:
This is the first post in this blog. To start, I will introduce the datasets that I used in my work. These are the datasets that have been used in the literature to train deepfakes models.
VIDTIMIT
Videos have a constant format
- width 512
- height 384
- fps 25
- frames (54-240)
Identities: 43 with each 10 videos.
Sentences: 10 sentences per identity (one per video) with the following 2 sentences for every person:
- “She had your dark suit in greasy wash water all year.”
- “Don’t ask me to carry an oily rag like that.”
The other 8 sentences are randomly selected from a corpus of 238 minus the previous two sentences.
Click me
```json [ "si1010", "si1024", "si1039", "si1084", "si1088", "si1099", "si1105", "si1116", "si1175", "si1178", "si1179", "si1199", "si1218", "si1230", "si1264", "si1265", "si1279", "si1303", "si1360", "si1364", "si1375", "si1386", "si1398", "si1400", "si1425", "si1454", "si1463", "si1469", "si1473", "si1474", "si1490", "si1502", "si1538", "si1539", "si1541", "si1542", "si1543", "si1544", "si1553", "si1555", "si1566", "si1573", "si1587", "si1624", "si1640", "si1653", "si1664", "si1669", "si1714", "si1718", "si1729", "si1735", "si1746", "si1808", "si1825", "si1829", "si1848", "si1894", "si1899", "si1909", "si1933", "si1988", "si2005", "si2016", "si2028", "si2030", "si2084", "si2093", "si2098", "si2099", "si2104", "si2120", "si2128", "si2149", "si2168", "si2169", "si2171", "si2172", "si2173", "si2183", "si2203", "si2222", "si2247", "si2255", "si2270", "si2284", "si2294", "si2299", "si458", "si469", "si475", "si522", "si545", "si548", "si549", "si565", "si569", "si634", "si635", "si639", "si649", "si673", "si728", "si730", "si734", "si745", "si756", "si768", "si770", "si818", "si824", "si833", "si838", "si839", "si844", "si860", "si869", "si908", "si909", "si911", "si912", "si913", "si923", "si943", "si992", "si995", "sx10", "sx100", "sx101", "sx102", "sx103", "sx104", "sx109", "sx11", "sx110", "sx113", "sx115", "sx118", "sx119", "sx1195", "sx12", "sx124", "sx126", "sx13", "sx133", "sx134", "sx138", "sx139", "sx14", "sx140", "sx1625", "sx184", "sx185", "sx188", "sx189", "sx19", "sx190", "sx191", "sx192", "sx193", "sx194", "sx199", "sx20", "sx200", "sx203", "sx205", "sx208", "sx209", "sx214", "sx216", "sx223", "sx224", "sx228", "sx229", "sx23", "sx230", "sx25", "sx274", "sx275", "sx278", "sx279", "sx28", "sx280", "sx281", "sx282", "sx283", "sx284", "sx289", "sx29", "sx290", "sx293", "sx295", "sx298", "sx299", "sx304", "sx306", "sx313", "sx314", "sx318", "sx319", "sx320", "sx34", "sx36", "sx364", "sx365", "sx368", "sx369", "sx370", "sx371", "sx372", "sx373", "sx374", "sx379", "sx380", "sx383", "sx385", "sx388", "sx389", "sx394", "sx396", "sx4", "sx403", "sx404", "sx408", "sx409", "sx410", "sx43", "sx44", "sx48", "sx49", "sx5", "sx50", "sx8", "sx9", "sx94", "sx95", "sx98", "sx99" ] ```total: 430 videos
label
- REAL 430 (100.00 %)
fold
- real 430 (100.00 %)
| label | fold | count | percent |
|---|---|---|---|
| 0 | real | 430 | 100 |
DeepfakeTIMIT
total: 1070 videos
label
- FAKE 640 (59.81 %)
- REAL 430 (40.19 %)
fold
- real 430 (40.19 %)
- fake_HQ 320 (29.91 %)
- fake_LQ 320 (29.91 %)
| label | fold | count | percent |
|---|---|---|---|
| 0 | real | 430 | 40.19 |
| 1 | fake_HQ | 320 | 29.91 |
| 1 | fake_LQ | 320 | 29.91 |
UADFV
total: 98 videos
generated with FakeAPP
label
- REAL 49 (50.00 %)
- FAKE 49 (50.00 %)
fold
- youtube 49 (50.00 %)
- FakeAPP 49 (50.00 %)
| label | fold | count | percent |
|---|---|---|---|
| 0 | youtube | 49 | 50 |
| 1 | FakeAPP | 49 | 50 |
FaceForensics++:
total: 6000 videos
label
- FAKE 5000 (83.33 %)
- REAL 1000 (16.67 %)
fold
- youtube 1000 (16.67 %)
- Deepfakes 1000 (16.67 %)
- Face2Face 1000 (16.67 %)
- FaceShifter 1000 (16.67 %)
- FaceSwap 1000 (16.67 %)
- NeuralTextures 1000 (16.67 %)
split
- train 4320 (72.00 %)
- test 840 (14.00 %)
- val 840 (14.00 %)
| label | fold | split | count | percent |
|---|---|---|---|---|
| 0 | youtube | test | 140 | 2.33 |
| 0 | youtube | train | 720 | 12 |
| 0 | youtube | val | 140 | 2.33 |
| 1 | Deepfakes | test | 140 | 2.33 |
| 1 | Deepfakes | train | 720 | 12 |
| 1 | Deepfakes | val | 140 | 2.33 |
| 1 | Face2Face | test | 140 | 2.33 |
| 1 | Face2Face | train | 720 | 12 |
| 1 | Face2Face | val | 140 | 2.33 |
| 1 | FaceShifter | test | 140 | 2.33 |
| 1 | FaceShifter | train | 720 | 12 |
| 1 | FaceShifter | val | 140 | 2.33 |
| 1 | FaceSwap | test | 140 | 2.33 |
| 1 | FaceSwap | train | 720 | 12 |
| 1 | FaceSwap | val | 140 | 2.33 |
| 1 | NeuralTextures | test | 140 | 2.33 |
| 1 | NeuralTextures | train | 720 | 12 |
| 1 | NeuralTextures | val | 140 | 2.33 |
Identities per split (json)
{ "train": [1,2,5,6,7,8,9,10,11,13,14,16,17,18,19,20,21,22,25,27,28,30,31,32,33,34,37,38,39,40,41,43,45,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,65,66,67,68,69,70,71,72,74,75,76,77,79,80,81,82,83,85,86,87,88,89,90,92,93,94,95,96,97,98,99,100,101,103,105,107,108,109,110,111,112,113,117,118,119,120,121,122,123,124,125,127,129,130,131,132,133,136,137,139,140,143,144,146,147,148,149,150,151,152,153,155,156,160,162,163,164,165,166,167,168,171,172,173,174,177,178,179,180,181,182,183,184,185,187,188,189,191,193,194,195,196,197,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,215,216,217,218,221,222,224,225,226,228,229,230,231,232,234,235,236,237,238,239,240,241,242,243,244,246,247,248,250,251,252,253,256,258,259,260,262,264,265,266,267,268,269,270,271,272,274,275,276,277,279,281,282,283,285,286,287,289,290,291,292,293,294,295,296,297,298,300,301,302,303,304,305,307,309,310,311,312,313,315,317,318,320,322,323,324,326,327,328,329,330,331,332,333,334,335,336,337,338,339,340,341,343,344,345,346,348,351,353,355,356,357,359,360,361,363,365,366,368,372,373,374,375,377,378,382,383,385,387,390,391,392,393,394,395,396,397,398,400,401,403,405,406,407,408,409,410,411,412,413,414,415,417,424,426,427,428,430,431,432,433,434,435,436,437,438,439,441,442,444,446,447,448,449,450,451,456,457,458,459,460,461,463,464,466,469,473,474,475,476,477,478,481,484,486,487,489,490,491,495,496,497,498,499,500,501,502,503,504,505,506,508,509,510,512,513,515,516,518,519,520,522,523,524,525,526,528,530,531,532,533,534,535,536,537,539,540,541,542,543,544,545,546,547,549,551,553,554,555,556,559,561,562,563,568,569,570,572,573,574,575,576,577,578,580,581,582,587,588,590,591,592,593,594,596,598,600,601,602,603,604,605,609,610,611,613,614,615,616,618,619,620,621,624,626,627,628,629,631,635,636,637,639,641,642,643,644,645,646,648,651,652,653,654,655,657,658,659,662,663,665,667,671,673,674,676,677,678,679,680,681,684,685,686,687,688,689,690,692,693,694,696,697,698,699,700,703,704,708,709,710,711,712,713,715,716,717,718,719,721,722,723,726,728,730,733,734,735,736,737,738,740,743,744,747,748,749,750,751,752,753,754,756,757,758,760,761,763,764,765,766,767,769,770,771,772,773,774,776,778,780,781,782,783,784,785,786,787,788,789,790,791,792,795,796,797,798,799,800,802,803,804,805,806,808,809,810,811,813,814,815,819,822,825,826,828,829,830,831,833,835,836,838,839,840,841,843,844,845,846,848,849,850,853,854,856,858,859,860,861,863,864,866,867,869,870,871,872,873,874,876,877,878,879,881,882,883,884,885,886,887,888,889,891,892,893,894,897,899,901,902,903,905,907,908,910,911,913,914,916,918,920,921,928,929,930,931,934,935,936,937,938,940,941,942,943,944,946,948,950,952,954,957,959,960,961,962,963,964,965,966,967,968,969,972,975,976,977,978,981,983,984,985,986,987,988,989,990,993,994,996,997,998,999], "test": [0,3,12,15,24,26,29,35,36,44,47,48,73,78,102,114,128,135,138,141,142,154,158,161,169,170,176,186,190,214,219,220,227,233,249,255,257,278,280,288,306,308,314,319,321,347,352,358,367,371,376,379,380,381,384,386,388,389,399,404,418,420,421,422,423,425,429,445,452,454,462,467,479,480,485,488,494,507,517,521,527,529,550,552,579,607,608,623,625,630,633,634,650,660,661,669,670,675,682,683,691,695,701,705,706,707,714,724,725,731,732,739,741,755,759,801,812,821,842,847,851,855,862,865,868,880,896,906,917,919,924,932,945,949,953,955,970,973,974,995], "val": [4,23,42,46,64,84,91,104,106,115,116,126,134,145,157,159,175,192,198,223,245,254,261,263,273,284,299,316,325,342,349,350,354,362,364,369,370,402,416,419,440,443,453,455,465,468,470,471,472,482,483,492,493,511,514,538,548,557,558,560,564,565,566,567,571,583,584,585,586,589,595,597,599,606,612,617,622,632,638,640,647,649,656,664,666,668,672,702,720,727,729,742,745,746,762,768,775,777,779,793,794,807,816,817,818,820,823,824,827,832,834,837,852,857,875,890,895,898,900,904,909,912,915,922,923,925,926,927,933,939,947,951,956,958,971,979,980,982,991,992] }
Celeb-DF-v1:
total: 1203 videos
label
- 1 795 (66.08 %)
- 0 408 (33.92 %)
fold
- Celeb-synthesis 795 (66.08 %)
- YouTube-real 250 (20.78 %)
- Celeb-real 158 (13.13 %)
split
- train 1103 (91.69 %)
- test 100 (8.31 %)
| label | fold | split | count | percent |
|---|---|---|---|---|
| 0 | Celeb-real | train | 150 | 12.47 |
| 0 | Celeb-real | test | 8 | 0.67 |
| 0 | YouTube-real | train | 220 | 18.29 |
| 0 | YouTube-real | test | 30 | 2.49 |
| 1 | Celeb-synthesis | train | 733 | 60.93 |
| 1 | Celeb-synthesis | test | 62 | 5.15 |
identities:
total = 15
[0, 1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 16, 17]
missing = 3
[5, 14, 15]
Celeb-DF-v2:
total: 6529 videos
label
- FAKE 5639 (86.37 %)
- REAL 890 (13.63 %)
fold
- Celeb-synthesis 5639 (86.37 %)
- Celeb-real 590 (9.04 %)
- YouTube-real 300 (4.59 %)
split
- train 6011 (92.07 %)
- test 518 (7.93 %)
| label | fold | split | count | percent |
|---|---|---|---|---|
| 0 | Celeb-real | train | 482 | 7.38 |
| 0 | Celeb-real | test | 108 | 1.65 |
| 0 | YouTube-real | train | 230 | 3.52 |
| 0 | YouTube-real | test | 70 | 1.07 |
| 1 | Celeb-synthesis | train | 5299 | 81.16 |
| 1 | Celeb-synthesis | test | 340 | 5.21 |
Identities
total = 15
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 16, 17, 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]
missing = 3
[14, 15, 18]
Celebrities
total: 59
Gender: male (56.8%) female (43.2%)
age:
=60 (8.5%); [50, 60] (30.5%); 40s (26.6%) 30s (28.0%) <30 (6.4%)
Ethnic groups: Asians (5.1%) African Americans (6.8%) Caucasians (88.1%)
Deep Fake Detection Dataset
DFD: The Google/Jigsaw DeepFake detection dataset
total: 3431 videos
label
- 1 3068 (89.42 %)
- 0 363 (10.58 %)
fold
- DeepfakeDetection 3068 (89.42 %)
- actors 363 (10.58 %)
scenes
- outside_talking_pan_laughing 291 (8.48 %)
- walk_down_hall_angry 276 (8.04 %)
- outside_talking_still_laughing 268 (7.81 %)
- kitchen_pan 263 (7.67 %)
- talking_against_wall 260 (7.58 %)
- podium_speech_happy 257 (7.49 %)
- exit_phone_room 246 (7.17 %)
- walking_outside_cafe_disgusted 239 (6.97 %)
- walking_down_street_outside_angry 229 (6.67 %)
- kitchen_still 225 (6.56 %)
- talking_angry_couch 198 (5.77 %)
- secret_conversation 182 (5.30 %)
- hugging_happy 180 (5.25 %)
- walking_and_outside_surprised 139 (4.05 %)
- walking_down_indoor_hall_disgust 121 (3.53 %)
- meeting_serious 57 (1.66 %)
| scene | count real | percent real | count fake | percent fake | |
|---|---|---|---|---|---|
| exit_phone_room | 28 | 0.82 | 218 | 6.35 | |
| hugging_happy | 16 | 0.47 | 164 | 4.78 | |
| kitchen_pan | 27 | 0.79 | 236 | 6.88 | |
| kitchen_still | 26 | 0.76 | 199 | 5.8 | |
| meeting_serious | 4 | 0.12 | 53 | 1.54 | |
| outside_talking_pan_laughing | 28 | 0.82 | 263 | 7.67 | |
| outside_talking_still_laughing | 27 | 0.79 | 241 | 7.02 | |
| podium_speech_happy | 28 | 0.82 | 229 | 6.67 | |
| secret_conversation | 18 | 0.52 | 164 | 4.78 | |
| talking_against_wall | 27 | 0.79 | 233 | 6.79 | |
| talking_angry_couch | 22 | 0.64 | 176 | 5.13 | |
| walk_down_hall_angry | 28 | 0.82 | 248 | 7.23 | |
| walking_and_outside_surprised | 15 | 0.44 | 124 | 3.61 | |
| walking_down_indoor_hall_disgust | 15 | 0.44 | 106 | 3.09 | |
| walking_down_street_outside_angry | 28 | 0.82 | 201 | 5.86 | |
| walking_outside_cafe_disgusted | 26 | 0.76 | 213 | 6.21 |
DFDC-preview:
total: 5250 videos
label
- FAKE 4119 (78.46 %)
- REAL 1131 (21.54 %)
fold
- method_A 3975 (75.71 %)
- original_videos 1131 (21.54 %)
- method_B 144 (2.74 %)
split
- train 4473 (85.20 %)
- test 777 (14.80 %)
| label | fold | split | count | percent |
|---|---|---|---|---|
| 1 | method_A | train | 3501 | 66.69 |
| 1 | method_A | test | 474 | 9.03 |
| 1 | method_B | train | 117 | 2.23 |
| 1 | method_B | test | 27 | 0.51 |
| 0 | original_videos | train | 855 | 16.29 |
| 0 | original_videos | test | 276 | 5.26 |
DFDC:
Facebook DeepFake Detection Challenge Dataset (DFDC)
total: 128154 videos
label
- REAL 23654 (81.54 %)
- FAKE 104500 (18.45 %)
split
- train 119154 (92.98 %)
- test 5000 (3.90 %)
- val 4000 (3.12 %)
| label | split | count | percent |
|---|---|---|---|
| 0 | test | 2500 | 1.95 |
| 0 | train | 19154 | 14.95 |
| 0 | val | 2000 | 1.56 |
| 1 | test | 2500 | 1.95 |
| 1 | train | 100000 | 78.03 |
| 1 | val | 2000 | 1.56 |
DeeperForensics-1.0
total: 59475 videos
label
- REAL 48475 (81.50 %)
- FAKE 11000 (18.50 %)
split
- train 41452 (69.70 %)
- test 12368 (20.80 %)
- val 5655 (9.51 %)
| label | split | count | percent |
|---|---|---|---|
| 0 | test | 10157 | 17.08 |
| 0 | train | 33719 | 56.69 |
| 0 | val | 4599 | 7.73 |
| 1 | test | 2211 | 3.72 |
| 1 | train | 7733 | 13 |
| 1 | val | 1056 | 1.78 |
Perturbation There are 7 types of distortions with 5 intensity (levels): (transmission errors, compression, etc.)
- CS: Color Staturation Change
- BW: Local Block Wise Distortion
- CC: Color Contrast Change
- GB: Gaussian blur
- GNC: White Gaussian Noise in Color Components
- JPEG: JPEG Compression
- VC: Video Compression Rate Change
Additionally, 3 distortion are created by mixing respectively 2, 3 and 4 of the above distortions, ie. {CS, BW}, {CS, BW, CC}, {CS, BW, CC, GB}.
Each forgery video is then augmented using:
- 1 original forgery
- 1 forgery postprocessed with the original frames by color matching, warping, affine transformation, etc.
- 1 random distortion at a random level
- 5 randomly choosen distortion at level-{1, 2, 3, 4, 5}
- 3 mix-{2, 3, 4} distortions
total: 11 videos per forgery
Identities: All the 11000 FAKE videos are created using source videos from DeeperForensics-1.0 dataset and the target videos are the 1000 pristine videos from FaceForensics++ dataset.
When keeping only the test 140 identities of FaceForensics++ used as source for the DeeperForensics-1.0 dataset, we get the following: label
- FAKE 1540 = 140 * 11 (100.00 %)
label
- 1 1540 (100.00 %)
split
- train 1100 (71.43 %)
- test 330 (21.43 %)
- val 110 (7.14 %)
manipulation
- end to end 140 (9.09 %)
- reenact postprocess 140 (9.09 %)
- end to end random level 140 (9.09 %)
- end to end level 1 140 (9.09 %)
- end to end level 2 140 (9.09 %)
- end to end level 3 140 (9.09 %)
- end to end level 4 140 (9.09 %)
- end to end level 5 140 (9.09 %)
- end to end mix 2 distortions 140 (9.09 %)
- end to end mix 3 distortions 140 (9.09 %)
- end to end mix 4 distortions 140 (9.09 %)
FFIW10K
total: 19976 videos
label
- REAL 9988 (50.00 %)
- FAKE 9988 (50.00 %)
fold
- web 19976 (100.00 %)
split
- train 16000 (80.10 %)
- test 3476 (17.40 %)
- val 500 (2.50 %)
| label | fold | split | count | percent |
|---|---|---|---|---|
| 0 | web | train | 8000 | 40.05 |
| 0 | web | val | 250 | 1.25 |
| 0 | web | test | 1738 | 8.7 |
| 1 | web | train | 8000 | 40.05 |
| 1 | web | val | 250 | 1.25 |
| 1 | web | test | 1738 | 8.7 |
Wilddeepfake
total: 7312 videos
label
- REAL 3805 (52.04 %)
- FAKE 3507 (47.96 %)
split
- train 6506 (88.98 %)
- test 806 (11.02 %)
| label | split | count | percent |
|---|---|---|---|
| 1 | test | 410 | 5.61 |
| 1 | train | 3097 | 42.36 |
| 0 | test | 396 | 5.42 |
| 0 | train | 3409 | 46.62 |
KoDF
total: 237824 videos
label
- 1 175660 (73.86 %)
- 0 62164 (26.14 %)
fold
- fo 61867 (26.01 %)
- crowdsourcing 52964 (22.27 %)
- dffs 37080 (15.59 %)
- dfl 34464 (14.49 %)
- fsgan 23888 (10.04 %)
- audio-driven 18361 (7.72 %)
- studio 9200 (3.87 %)
sex
- F 130502 (54.87 %)
- M 107322 (45.13 %)
| label | fold | sex | count | percent |
|---|---|---|---|---|
| 0 | crowdsourcing | F | 27023 | 11.36 |
| 0 | crowdsourcing | M | 25941 | 10.91 |
| 0 | studio | F | 4600 | 1.93 |
| 0 | studio | M | 4600 | 1.93 |
| 1 | fo | F | 32615 | 13.71 |
| 1 | fo | M | 29252 | 12.3 |
| 1 | audio-driven | F | 6870 | 2.89 |
| 1 | audio-driven | M | 11491 | 4.83 |
| 1 | dfl | F | 21910 | 9.21 |
| 1 | dfl | M | 12554 | 5.28 |
| 1 | dffs | F | 22575 | 9.49 |
| 1 | dffs | M | 14505 | 6.1 |
| 1 | fsgan | F | 14909 | 6.27 |
| 1 | fsgan | M | 8979 | 3.78 |
Acronyms:
- fo: FOMM
- dffs: FaceSwap
- dfl: DeepFaceLab
- fsgan: FSGAN
- audio-driven: AudioDriven : {ATFHP, Wav2Lip}
| Age | Count | % |
|---|---|---|
| ∼19 | 5 | 1.24 |
| 20∼29 | 205 | 50.87 |
| 30∼39 | 106 | 26.30 |
| 40∼49 | 61 | 15.14 |
| 50∼59 | 19 | 4.71 |
| 60∼ | 7 | 1.74 |
| Sex | Count | % |
|---|---|---|
| Female | 205 | 50.87 |
| Male | 198 | 49.13 |
| Location | Count | % |
|---|---|---|
| Crowdsourcing | 353 | 87.59 |
| Studio | 50 | 12.41 |
CDDB
without ProGAN
total: 81847 videos label
- 1 40925 (50.00 %)
- 0 40922 (50.00 %)
family
- non-gan 37098 (45.33 %)
- gan 34878 (42.61 %)
- unknown 9871 (12.06 %)
method
- glow 12000 (14.66 %)
- stargan_gf 12000 (14.66 %)
- crn 10210 (12.47 %)
- imle 10208 (12.47 %)
- stylegan 9582 (11.71 %)
- wild 8271 (10.11 %)
- gaugan 8000 (9.77 %)
- deepfake 4330 (5.29 %)
- biggan 3200 (3.91 %)
- cyclegan 2096 (2.56 %)
- whichfaceisreal 1600 (1.95 %)
- san 350 (0.43 %)
| label | method | count | percent |
|---|---|---|---|
| 0 | biggan | 1600 | 1.95 |
| 0 | crn | 5105 | 6.24 |
| 0 | cyclegan | 1048 | 1.28 |
| 0 | deepfake | 2165 | 2.65 |
| 0 | gaugan | 4000 | 4.89 |
| 0 | glow | 6000 | 7.33 |
| 0 | imle | 5104 | 6.24 |
| 0 | san | 175 | 0.21 |
| 0 | stargan_gf | 6000 | 7.33 |
| 0 | stylegan | 4791 | 5.85 |
| 0 | whichfaceisreal | 800 | 0.98 |
| 0 | wild | 4134 | 5.05 |
| 1 | biggan | 1600 | 1.95 |
| 1 | crn | 5105 | 6.24 |
| 1 | cyclegan | 1048 | 1.28 |
| 1 | deepfake | 2165 | 2.65 |
| 1 | gaugan | 4000 | 4.89 |
| 1 | glow | 6000 | 7.33 |
| 1 | imle | 5104 | 6.24 |
| 1 | san | 175 | 0.21 |
| 1 | stargan_gf | 6000 | 7.33 |
| 1 | stylegan | 4791 | 5.85 |
| 1 | whichfaceisreal | 800 | 0.98 |
| 1 | wild | 4137 | 5.05 |