Dataset Viewer
Auto-converted to Parquet Duplicate
image
imagewidth (px)
1.05k
1.92k
scene
stringclasses
20 values
category
stringclasses
2 values
image_name
stringlengths
9
15
camera_model
stringclasses
1 value
width
int64
1.05k
1.92k
height
int64
715
1.91k
fx
float64
918
1.79k
fy
float64
918
1.79k
cx
float64
523
960
cy
float64
358
957
qw
float64
-0.51
1
qx
float64
-0.36
0.2
qy
float64
-0.96
1
qz
float64
-0.62
0.73
tx
float64
-5.3
6.22
ty
float64
-3.58
2.88
tz
float64
-6.4
6.98
babies_cerny
statues
100040.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.956236
-0.012717
0.287969
0.050256
0.265093
-0.21017
2.284572
babies_cerny
statues
100005.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.752893
0.016287
-0.650294
-0.100019
1.594632
-0.726518
5.20006
babies_cerny
statues
200004.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.76732
-0.021575
0.637488
0.066052
-1.625857
-0.326388
2.115099
babies_cerny
statues
200012.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.663839
-0.019929
0.740639
0.101852
-2.244026
-0.638683
2.776305
babies_cerny
statues
100037.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.976914
-0.00302
0.210042
0.038897
0.520528
-0.208248
2.013154
babies_cerny
statues
100017.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.932208
0.018496
-0.357451
-0.053621
1.720448
-0.437226
3.891004
babies_cerny
statues
200008.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.717397
-0.027967
0.690533
0.087885
-2.066008
-0.50212
2.489799
babies_cerny
statues
200011.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.668659
-0.019252
0.736951
0.097095
-2.152379
-0.589271
2.797064
babies_cerny
statues
100047.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.902983
-0.022079
0.421503
0.080439
-0.475351
-0.335848
2.323268
babies_cerny
statues
100022.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.978484
0.021965
-0.204498
-0.016326
1.7069
-0.293956
3.167401
babies_cerny
statues
200024.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.505837
-0.020669
0.853145
0.125877
-2.342266
-0.974083
4.04511
babies_cerny
statues
100006.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.771437
0.015258
-0.628194
-0.100124
1.547661
-0.705703
5.123274
babies_cerny
statues
200016.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.613745
-0.00916
0.782552
0.104136
-2.196477
-0.761802
3.515089
babies_cerny
statues
200019.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.586089
-0.008462
0.801985
0.115102
-2.396625
-0.895029
3.776688
babies_cerny
statues
100042.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.943518
-0.008229
0.325437
0.061623
0.054837
-0.261541
2.268724
babies_cerny
statues
100015.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.910168
0.021822
-0.410423
-0.051689
1.722564
-0.411663
4.017295
babies_cerny
statues
100016.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.923073
0.017208
-0.380594
-0.052805
1.698067
-0.392759
3.954608
babies_cerny
statues
100035.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.985393
0.00139
0.166858
0.034008
0.788097
-0.178251
1.936402
babies_cerny
statues
100012.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.873278
0.027884
-0.480908
-0.073049
1.784669
-0.560602
4.346494
babies_cerny
statues
100041.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.949529
-0.010737
0.308073
0.058063
0.164745
-0.234869
2.253736
babies_cerny
statues
200028.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.429845
-0.019353
0.893203
0.130565
-2.307291
-1.050564
4.509325
babies_cerny
statues
100021.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.970452
0.024537
-0.238834
-0.024051
1.659943
-0.335973
3.307251
babies_cerny
statues
200017.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.602525
-0.006101
0.790653
0.108604
-2.239714
-0.816093
3.703694
babies_cerny
statues
100019.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.953464
0.015105
-0.298054
-0.042927
1.729411
-0.347265
3.571696
babies_cerny
statues
100009.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.826441
0.018225
-0.555819
-0.087912
1.715065
-0.611752
4.71691
babies_cerny
statues
100029.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.999945
0.010513
0.000025
0.000601
1.288177
-0.214849
2.43281
babies_cerny
statues
100023.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.984554
0.01477
-0.173945
-0.013338
1.588811
-0.261448
3.161282
babies_cerny
statues
100038.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.972282
-0.008295
0.230223
0.039954
0.406737
-0.196681
2.061734
babies_cerny
statues
100007.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.788479
0.011957
-0.60728
-0.096792
1.583287
-0.678291
5.014043
babies_cerny
statues
200020.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.563507
-0.012103
0.817516
0.118244
-2.308714
-0.913108
3.874509
babies_cerny
statues
100011.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.855827
0.027845
-0.509802
-0.082986
1.685713
-0.634264
4.481666
babies_cerny
statues
100020.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.961938
0.018359
-0.270436
-0.034668
1.661014
-0.339337
3.493658
babies_cerny
statues
100033.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.993393
0.00508
0.112514
0.022012
0.948179
-0.18896
2.152907
babies_cerny
statues
100002.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.688893
0.001038
-0.718397
-0.096598
1.572534
-0.599876
5.807356
babies_cerny
statues
100008.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.80571
0.013465
-0.585069
-0.091353
1.627008
-0.629783
4.861354
babies_cerny
statues
100032.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.996367
0.005545
0.084105
0.012179
1.07803
-0.201727
2.195821
babies_cerny
statues
100013.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.887613
0.027443
-0.45523
-0.064472
1.757123
-0.511374
4.284851
babies_cerny
statues
100050.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.869996
-0.02966
0.485887
0.078367
-0.660352
-0.323211
2.459506
babies_cerny
statues
100025.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.994558
0.01842
-0.102292
-0.007113
1.566842
-0.289899
2.906249
babies_cerny
statues
200022.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.533127
-0.013524
0.836918
0.123126
-2.307705
-0.961574
3.976185
babies_cerny
statues
200013.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.646465
-0.021648
0.754712
0.109654
-2.258457
-0.69526
2.863913
babies_cerny
statues
200009.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.700213
-0.028905
0.707291
0.092766
-2.065425
-0.533631
2.570575
babies_cerny
statues
100034.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.989706
0.003783
0.139919
0.029836
0.86542
-0.189108
2.016672
babies_cerny
statues
200021.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.545084
-0.011939
0.829372
0.121999
-2.301259
-0.950163
3.967689
babies_cerny
statues
100010.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.842501
0.023573
-0.531056
-0.087273
1.714673
-0.650062
4.624758
babies_cerny
statues
200014.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.628084
-0.019374
0.770389
0.107873
-2.263742
-0.69162
3.024723
babies_cerny
statues
200026.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.472815
-0.021227
0.871568
0.127924
-2.331722
-1.007363
4.218996
babies_cerny
statues
100039.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.96427
-0.010694
0.261305
0.042303
0.312537
-0.186385
2.196461
babies_cerny
statues
200005.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.751831
-0.030905
0.65442
0.074363
-1.703978
-0.359661
2.137335
babies_cerny
statues
100051.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.859483
-0.033554
0.503464
0.08177
-0.734962
-0.338752
2.462761
babies_cerny
statues
200027.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.451317
-0.021872
0.882912
0.127678
-2.292739
-1.003867
4.316567
babies_cerny
statues
200003.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.777489
-0.017162
0.626111
0.05659
-1.526813
-0.277874
2.124134
babies_cerny
statues
100026.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.997214
0.018499
-0.072143
-0.00416
1.484697
-0.281742
2.791128
babies_cerny
statues
100046.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.910818
-0.019235
0.405877
0.072831
-0.410374
-0.306373
2.339517
babies_cerny
statues
200018.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.596901
-0.004001
0.794294
0.113094
-2.369953
-0.883173
3.767571
babies_cerny
statues
100028.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.999546
0.012934
-0.027227
-0.000175
1.382145
-0.236421
2.539392
babies_cerny
statues
200025.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.488538
-0.020575
0.86286
0.127987
-2.337747
-1.003859
4.165704
babies_cerny
statues
100048.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.89318
-0.029479
0.440738
0.084324
-0.493687
-0.316579
2.310755
babies_cerny
statues
100004.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.735418
0.011162
-0.669947
-0.101033
1.616791
-0.708711
5.351904
babies_cerny
statues
100018.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.942322
0.01739
-0.33044
-0.050367
1.717416
-0.403226
3.726955
babies_cerny
statues
200007.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.725612
-0.026282
0.68216
0.086338
-1.981373
-0.489926
2.497063
babies_cerny
statues
100027.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.998665
0.01572
-0.049197
-0.000281
1.436455
-0.253989
2.685259
babies_cerny
statues
100031.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.998599
0.007487
0.051602
0.008976
1.102411
-0.214971
2.296185
babies_cerny
statues
100024.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.990072
0.014913
-0.139359
-0.010707
1.563878
-0.289934
3.051827
babies_cerny
statues
200006.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.736434
-0.030154
0.670818
0.082207
-1.882516
-0.446083
2.354201
babies_cerny
statues
200001.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.787751
-0.014396
0.614184
0.044943
-1.276038
-0.227295
2.133637
babies_cerny
statues
100003.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.712529
0.003919
-0.693989
-0.103282
1.584815
-0.666963
5.563034
babies_cerny
statues
200010.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.677225
-0.025074
0.728892
0.097232
-2.07927
-0.57141
2.726554
babies_cerny
statues
100030.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.999638
0.008628
0.024718
0.006122
1.206059
-0.22193
2.413419
babies_cerny
statues
200015.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.619909
-0.013896
0.777704
0.103425
-2.248085
-0.729766
3.265029
babies_cerny
statues
100036.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.980601
0.000989
0.192358
0.037658
0.618691
-0.200636
1.987142
babies_cerny
statues
200023.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.522668
-0.017637
0.843319
0.123774
-2.344669
-0.955071
3.960081
babies_cerny
statues
100014.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.897645
0.028261
-0.436253
-0.055831
1.694941
-0.470616
4.150006
babies_cerny
statues
100043.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.937546
-0.013853
0.341471
0.064903
-0.070738
-0.270467
2.344037
babies_cerny
statues
100044.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.930071
-0.020859
0.360734
0.066355
-0.180465
-0.257209
2.297122
babies_cerny
statues
100001.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.67946
0.005608
-0.728544
-0.086752
1.573879
-0.541241
6.171172
babies_cerny
statues
200002.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.783027
-0.015952
0.619898
0.048387
-1.395384
-0.231808
2.118251
babies_cerny
statues
100045.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.919926
-0.02195
0.385363
0.068912
-0.229286
-0.268017
2.286827
babies_cerny
statues
100049.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.880826
-0.03073
0.465536
0.080487
-0.553848
-0.303376
2.340822
babies_cerny
statues
100052.png
PINHOLE
1,069
1,902
1,771.252334
1,771.252334
534.5
951
0.846181
-0.034883
0.52453
0.087345
-0.75114
-0.357569
2.497841
memorial_victims_of_communism
statues
100040.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.992399
0.003985
-0.119736
0.028128
-0.679856
-1.299675
4.337225
memorial_victims_of_communism
statues
100005.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
-0.344075
-0.082389
0.891284
0.283615
2.24066
-0.480365
3.929106
memorial_victims_of_communism
statues
200004.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.96439
-0.056121
-0.258198
-0.011705
-0.348606
-0.883562
4.020152
memorial_victims_of_communism
statues
200012.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.985824
-0.001639
-0.167656
0.006377
-1.186348
-1.269438
4.360558
memorial_victims_of_communism
statues
200033.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.95778
-0.0518
0.27741
0.054941
-0.521325
-0.946771
3.577122
memorial_victims_of_communism
statues
100037.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.97124
0.034148
-0.234731
0.020685
-1.277737
-1.489777
3.235833
memorial_victims_of_communism
statues
100017.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
-0.323481
-0.059623
0.896742
0.296072
0.353871
-0.381908
1.728941
memorial_victims_of_communism
statues
200081.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.952439
-0.128415
-0.274742
0.029773
3.225909
0.64613
-3.313888
memorial_victims_of_communism
statues
200038.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.979517
-0.10037
0.171662
0.031689
-1.399962
-0.690755
2.183694
memorial_victims_of_communism
statues
200079.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.949868
-0.117398
-0.287582
0.035569
3.088923
0.741323
-2.843329
memorial_victims_of_communism
statues
200084.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.925569
-0.114393
-0.360184
0.022426
4.080974
0.968846
-3.143081
memorial_victims_of_communism
statues
200089.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.769781
-0.041725
-0.636941
-0.001404
6.05971
2.031509
-0.20813
memorial_victims_of_communism
statues
200045.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.96293
-0.087049
0.252708
0.036417
-1.813276
-0.439799
1.534131
memorial_victims_of_communism
statues
200051.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.955661
-0.076255
0.282345
0.034334
-2.319233
-0.114672
0.394295
memorial_victims_of_communism
statues
200008.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.962463
-0.019433
-0.270714
0.001081
-1.288319
-1.170408
4.307036
memorial_victims_of_communism
statues
200011.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.982284
-0.006296
-0.18729
-0.000353
-1.163349
-1.22509
4.365819
memorial_victims_of_communism
statues
200061.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.987764
-0.148234
-0.043942
0.020443
0.49273
0.127191
-1.343809
memorial_victims_of_communism
statues
100022.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
-0.369992
-0.025868
0.896497
0.242343
-0.262161
-0.156852
0.819732
memorial_victims_of_communism
statues
200035.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.971955
-0.068465
0.222707
0.031889
-0.982605
-0.826166
3.035456
memorial_victims_of_communism
statues
200024.png
PINHOLE
1,060
1,886
1,752.330064
1,752.330064
530
943
0.987453
-0.000394
0.150607
0.047469
-0.28838
-1.290317
4.525128
End of preview. Expand in Data Studio

The D2P dataset

The D2P dataset is a dataset based on the Depth2Pose monocular depth estimation benchmark, a pose-based evaluation of MDEs without ground-truth depth. The dataset contains challenging scenes beyond the distribution of common training data, together with a simple and extensible evaluation framework, presented on the github page. The scenes are divided into two categories: statues and vegetation. Undistorted images and reconstructions in standard colmap format is provided for each scene, together with a list of image pairs used for the evaluation.

Paper | GitHub | Webpage

Dataset Structure

d2p_dataset                           
β”œβ”€β”€ statues/                           
β”‚   β”œβ”€β”€ scene1/                          
β”‚   β”‚   β”œβ”€β”€ images/                       
β”‚   β”‚   β”‚   β”œβ”€β”€ img1.png
β”‚   β”‚   β”‚   β”œβ”€β”€ img2.png   
β”‚   β”‚   β”‚   └── ... 
β”‚   β”‚   β”œβ”€β”€ sparse/
β”‚   β”‚   β”‚   β”œβ”€β”€ cameras.txt
β”‚   β”‚   β”‚   β”œβ”€β”€ frames.txt
β”‚   β”‚   β”‚   β”œβ”€β”€ images.txt
β”‚   β”‚   β”‚   β”œβ”€β”€ points3D.txt
β”‚   β”‚   β”‚   └── rigs.txt
β”‚   β”‚   β”œβ”€β”€ scene1_image_list.txt
β”‚   β”‚   └── scene1_image_pairs.txt
β”‚   β”œβ”€β”€ scene2/ 
β”‚   β”‚   └── ... 
β”‚   └── ... 
└── vegetation/

Dataset Fields

Each scene contains:

  • images/: RGB images
  • sparse/: COLMAP reconstruction files:
    • camera parameters
    • frames
    • image poses
    • 3D points
    • rigs
  • scene1_image_list.txt: List of all images used for the benchmark, found in the images/ folder
  • scene1_image_pairs.txt: List of all image pairs used for the benchmark, for which relative pose is evaluated

Pre-extracted benchmark data

The folder d2p_extracted/ contains the ground truth and correspondences used in the paper, in the format read directly by the evaluation code. With it, the pair extraction (datasets/colmap.py) and matching (matchers/) steps can be skipped and only depth inference and pose evaluation need to be run.

d2p_extracted
β”œβ”€β”€ SHA256SUMS
β”œβ”€β”€ statues/
β”‚   β”œβ”€β”€ scene1/
β”‚   β”‚   β”œβ”€β”€ scene1.h5                          # ground-truth poses and intrinsics
β”‚   β”‚   β”œβ”€β”€ scene1_image_list.txt              # images used in the benchmark
β”‚   β”‚   β”œβ”€β”€ scene1_image_pairs.txt             # the 250 evaluated image pairs
β”‚   β”‚   β”œβ”€β”€ scene1_loma_2048_noresize.h5       # LoMa matches
β”‚   β”‚   └── scene1_splg_2048_noresize.h5       # SuperPoint + LightGlue matches
β”‚   └── ...
└── vegetation/

All files are HDF5 with one dataset per key. Image names are paths relative to the scene folder in d2p_dataset/, written with a Windows separator (e.g. images\00001.png), exactly as in the *_image_list.txt and *_image_pairs.txt files. Every file contains an empty completed group (marks a finished extraction) and a metadata group (git commit and machine info of the run that produced it).

Ground truth (<scene>.h5), for every image <name> in <scene>_image_list.txt:

key shape content
<name>_R 3x3 world-to-camera rotation (COLMAP convention, x_cam = R X + T)
<name>_T 3 world-to-camera translation
<name>_K 3x3 intrinsic matrix (pinhole, undistorted images)
<name>_size 2 image size (width, height)
<name>_K_orig, <name>_size_orig same as above for the original resolution (identical for D2P)

These values are identical to the COLMAP models in d2p_dataset/*/*/sparse/.

Matches (<scene>_<matcher>.h5), for every line <name1>,<name2> in <scene>_image_pairs.txt, the key <name1>-<name2> holds an N x C float32 array with one row per match; the first four columns are the pixel coordinates x1, y1, x2, y2 in the full-resolution images.

file matcher columns
*_loma_2048_noresize.h5 LoMa, 2048 keypoints, no resizing x1, y1, x2, y2
*_splg_2048_noresize.h5 SuperPoint + LightGlue, 2048 keypoints, no resizing x1, y1, x2, y2, score

In the evaluation code these are selected with --matches loma_2048_noresize or --matches splg_2048_noresize.

Direct Use

Benchmarking monocular depth estimators. For the current leaderboard, see the Depth2Pose webpage

Load with πŸ€— Datasets

from datasets import load_dataset

ds = load_dataset("floodgab/d2p_dataset")
print(ds["evaluation"][0])

Loading Example

To download the Depth2Pose dataset:

from huggingface_hub import snapshot_download

path = snapshot_download(
    repo_id="floodgab/d2p_dataset",
    repo_type="dataset",
    local_dir="./d2p_dataset"  # Optional: saves directly to a folder you can easily find
)

To download only the files needed by the evaluation code (images, COLMAP models and the pre-extracted ground truth and matches, without the parquet copy in data/)

from huggingface_hub import snapshot_download

path = snapshot_download(
    "floodgab/d2p_dataset",
    repo_type="dataset",
    local_dir="d2p",
    allow_patterns=["d2p_dataset/*", "d2p_extracted/*"]
)

Citation

If you use Depth2Pose in your research or find our work helpful, please cite:

@inproceedings{kocur2026depth2pose,
  title     = {{Depth2Pose}: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth},
  author    = {Kocur, Viktor and Aung, Sithu and Flood, Gabrielle and Ding, Yaqing and Bujnak, Lukas and Sattler, Torsten and Kukelova, Zuzana},
  booktitle = {Advances in Neural Information Processing Systems},
  year      = {2026}
}
Downloads last month
218

Paper for floodgab/d2p_dataset