Detailled semantic labeling (2D) result


Name J. Sherrah (DFCN IRRG nDSM atrous CRF)
Affiliation DST Group, Australia
Abbreviation DST_6
Strategy (u)nsupervised, (s)upervised, (h)ybrid s


Overall statistics, reference set: full_reference


Number of non processed tiles by this participant: 0
↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.926
0.017
0.028
0.020
0.003
0.007
building
0.018
0.966
0.009
0.003
0.000
0.004
low_veg
0.058
0.013
0.853
0.067
0.000
0.009
tree
0.037
0.006
0.115
0.838
0.001
0.002
car
0.146
0.009
0.004
0.025
0.807
0.008
clutter
0.315
0.127
0.139
0.024
0.005
0.390
Precision/Correctness
0.879
0.938
0.829
0.863
0.922
0.765
Recall/Completeness
0.926
0.966
0.853
0.838
0.807
0.390
F1
0.902
0.952
0.841
0.850
0.861
0.516

Overall accuracy 0.878

Overall statistics, reference set: no_boundary


Number of non processed tiles by this participant: 0
↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.944
0.013
0.022
0.014
0.001
0.006
building
0.014
0.973
0.007
0.002
0.000
0.003
low_veg
0.045
0.008
0.882
0.057
0.000
0.008
tree
0.028
0.005
0.100
0.864
0.001
0.002
car
0.051
0.010
0.002
0.019
0.908
0.009
clutter
0.310
0.111
0.123
0.020
0.005
0.430
Precision/Correctness
0.904
0.956
0.854
0.891
0.961
0.788
Recall/Completeness
0.944
0.973
0.882
0.864
0.908
0.430
F1
0.924
0.964
0.868
0.877
0.934
0.556

Overall accuracy 0.902
=============================================================

Tile top_potsdam_2_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9017
0.0322
0.0286
0.0243
0.0024
0.0108
building
0.0375
0.9325
0.0196
0.0081
0.0000
0.0023
low_veg
0.0531
0.0255
0.8549
0.0588
0.0001
0.0076
tree
0.0272
0.0103
0.1325
0.8268
0.0009
0.0022
car
0.1251
0.0477
0.0034
0.0228
0.7893
0.0116
clutter
0.1606
0.2770
0.1471
0.0467
0.0040
0.3647
Precision/Correctness
0.867
0.878
0.854
0.879
0.909
0.538
Recall/Completeness
0.902
0.932
0.855
0.827
0.789
0.365
F1
0.884
0.904
0.854
0.852
0.845
0.435

Overall accuracy 0.864

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_2_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9270
0.0259
0.0208
0.0166
0.0005
0.0092
building
0.0331
0.9423
0.0160
0.0067
0.0000
0.0019
low_veg
0.0452
0.0218
0.8805
0.0461
0.0000
0.0064
tree
0.0198
0.0085
0.1123
0.8571
0.0006
0.0018
car
0.0505
0.0532
0.0011
0.0161
0.8655
0.0136
clutter
0.1246
0.2848
0.1117
0.0365
0.0048
0.4376
Precision/Correctness
0.894
0.902
0.881
0.909
0.961
0.551
Recall/Completeness
0.927
0.942
0.881
0.857
0.865
0.438
F1
0.910
0.922
0.881
0.882
0.911
0.488

Overall accuracy 0.892

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_2_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9059
0.0099
0.0611
0.0191
0.0020
0.0020
building
0.0043
0.9827
0.0060
0.0044
0.0000
0.0026
low_veg
0.0265
0.0048
0.9067
0.0602
0.0000
0.0019
tree
0.0146
0.0048
0.1168
0.8620
0.0006
0.0012
car
0.1245
0.0089
0.0062
0.0283
0.8253
0.0068
clutter
0.1154
0.3441
0.1423
0.0747
0.0000
0.3235
Precision/Correctness
0.888
0.912
0.878
0.915
0.905
0.404
Recall/Completeness
0.906
0.983
0.907
0.862
0.825
0.323
F1
0.897
0.946
0.892
0.888
0.863
0.359

Overall accuracy 0.893

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_2_14_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9334
0.0057
0.0459
0.0128
0.0006
0.0016
building
0.0017
0.9915
0.0025
0.0024
0.0000
0.0019
low_veg
0.0208
0.0024
0.9252
0.0501
0.0000
0.0015
tree
0.0107
0.0040
0.1025
0.8814
0.0004
0.0010
car
0.0339
0.0100
0.0014
0.0234
0.9236
0.0078
clutter
0.0723
0.3968
0.0944
0.0760
0.0000
0.3605
Precision/Correctness
0.915
0.941
0.897
0.932
0.950
0.399
Recall/Completeness
0.933
0.992
0.925
0.881
0.924
0.360
F1
0.924
0.965
0.911
0.906
0.937
0.379

Overall accuracy 0.914

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_3_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9013
0.0232
0.0360
0.0286
0.0038
0.0070
building
0.0097
0.9811
0.0042
0.0037
0.0000
0.0013
low_veg
0.0632
0.0257
0.8358
0.0693
0.0001
0.0058
tree
0.0254
0.0076
0.0948
0.8697
0.0011
0.0014
car
0.1311
0.0044
0.0021
0.0369
0.8161
0.0094
clutter
0.3343
0.1921
0.2505
0.0288
0.0051
0.1892
Precision/Correctness
0.873
0.930
0.836
0.888
0.910
0.536
Recall/Completeness
0.901
0.981
0.836
0.870
0.816
0.189
F1
0.887
0.955
0.836
0.879
0.861
0.280

Overall accuracy 0.880

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_3_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9295
0.0159
0.0272
0.0199
0.0008
0.0067
building
0.0068
0.9879
0.0019
0.0025
0.0000
0.0009
low_veg
0.0506
0.0147
0.8734
0.0563
0.0001
0.0048
tree
0.0182
0.0059
0.0799
0.8942
0.0008
0.0011
car
0.0397
0.0047
0.0003
0.0263
0.9185
0.0104
clutter
0.3491
0.1933
0.2127
0.0192
0.0063
0.2194
Precision/Correctness
0.905
0.954
0.867
0.917
0.960
0.545
Recall/Completeness
0.929
0.988
0.873
0.894
0.918
0.219
F1
0.917
0.971
0.870
0.906
0.939
0.313

Overall accuracy 0.910

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_3_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.8867
0.0246
0.0481
0.0325
0.0036
0.0045
building
0.0101
0.9815
0.0046
0.0026
0.0002
0.0010
low_veg
0.0367
0.0133
0.8866
0.0584
0.0000
0.0049
tree
0.0191
0.0054
0.1502
0.8211
0.0006
0.0035
car
0.1305
0.0168
0.0022
0.0505
0.7899
0.0102
clutter
0.2443
0.2033
0.1826
0.0332
0.0025
0.3341
Precision/Correctness
0.877
0.929
0.813
0.895
0.903
0.740
Recall/Completeness
0.887
0.982
0.887
0.821
0.790
0.334
F1
0.882
0.954
0.848
0.856
0.843
0.460

Overall accuracy 0.873

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_3_14_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9179
0.0176
0.0357
0.0243
0.0009
0.0037
building
0.0081
0.9876
0.0019
0.0016
0.0001
0.0007
low_veg
0.0255
0.0067
0.9169
0.0471
0.0000
0.0038
tree
0.0145
0.0041
0.1335
0.8446
0.0005
0.0028
car
0.0467
0.0183
0.0012
0.0376
0.8860
0.0103
clutter
0.2314
0.1974
0.1515
0.0265
0.0022
0.3910
Precision/Correctness
0.910
0.951
0.839
0.921
0.956
0.782
Recall/Completeness
0.918
0.988
0.917
0.845
0.886
0.391
F1
0.914
0.969
0.876
0.881
0.920
0.521

Overall accuracy 0.901

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9331
0.0088
0.0277
0.0266
0.0022
0.0016
building
0.0141
0.9729
0.0064
0.0044
0.0000
0.0021
low_veg
0.0536
0.0116
0.8316
0.0969
0.0002
0.0060
tree
0.0334
0.0036
0.0663
0.8946
0.0011
0.0012
car
0.1632
0.0019
0.0056
0.0420
0.7867
0.0006
clutter
0.5417
0.1206
0.1620
0.0704
0.0035
0.1018
Precision/Correctness
0.832
0.961
0.800
0.796
0.940
0.758
Recall/Completeness
0.933
0.973
0.832
0.895
0.787
0.102
F1
0.880
0.967
0.816
0.843
0.857
0.179

Overall accuracy 0.867

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9523
0.0050
0.0216
0.0195
0.0004
0.0011
building
0.0088
0.9833
0.0030
0.0032
0.0000
0.0017
low_veg
0.0400
0.0052
0.8632
0.0862
0.0000
0.0053
tree
0.0226
0.0022
0.0523
0.9216
0.0007
0.0008
car
0.0613
0.0010
0.0020
0.0326
0.9021
0.0009
clutter
0.5662
0.1077
0.1458
0.0631
0.0039
0.1133
Precision/Correctness
0.860
0.974
0.838
0.833
0.972
0.793
Recall/Completeness
0.952
0.983
0.863
0.922
0.902
0.113
F1
0.904
0.979
0.850
0.875
0.936
0.198

Overall accuracy 0.895

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.8895
0.0221
0.0618
0.0150
0.0032
0.0085
building
0.0175
0.9684
0.0037
0.0057
0.0000
0.0047
low_veg
0.0887
0.0162
0.8140
0.0573
0.0002
0.0235
tree
0.0314
0.0052
0.1560
0.8016
0.0012
0.0045
car
0.1389
0.0110
0.0043
0.0373
0.7940
0.0145
clutter
0.5132
0.0544
0.1417
0.0107
0.0006
0.2794
Precision/Correctness
0.730
0.924
0.757
0.883
0.941
0.740
Recall/Completeness
0.889
0.968
0.814
0.802
0.794
0.279
F1
0.802
0.946
0.785
0.840
0.861
0.406

Overall accuracy 0.809

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_14_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9095
0.0170
0.0556
0.0099
0.0005
0.0075
building
0.0140
0.9758
0.0017
0.0046
0.0000
0.0039
low_veg
0.0759
0.0104
0.8406
0.0497
0.0000
0.0234
tree
0.0232
0.0037
0.1364
0.8321
0.0009
0.0037
car
0.0558
0.0120
0.0014
0.0301
0.8850
0.0156
clutter
0.5281
0.0504
0.1281
0.0080
0.0004
0.2851
Precision/Correctness
0.745
0.942
0.780
0.908
0.981
0.762
Recall/Completeness
0.910
0.976
0.841
0.832
0.885
0.285
F1
0.819
0.959
0.809
0.868
0.931
0.415

Overall accuracy 0.830

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_15_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9376
0.0152
0.0356
0.0067
0.0024
0.0025
building
0.0095
0.9826
0.0040
0.0016
0.0000
0.0023
low_veg
0.0645
0.0260
0.8712
0.0335
0.0004
0.0044
tree
0.0726
0.0098
0.1491
0.7613
0.0048
0.0024
car
0.1702
0.0069
0.0076
0.0149
0.7894
0.0110
clutter
0.3985
0.2330
0.1394
0.0144
0.0031
0.2115
Precision/Correctness
0.893
0.945
0.795
0.915
0.923
0.682
Recall/Completeness
0.938
0.983
0.871
0.761
0.789
0.212
F1
0.915
0.964
0.832
0.831
0.851
0.323

Overall accuracy 0.893

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_4_15_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9553
0.0088
0.0298
0.0041
0.0005
0.0015
building
0.0066
0.9882
0.0021
0.0011
0.0000
0.0020
low_veg
0.0426
0.0159
0.9120
0.0261
0.0001
0.0034
tree
0.0557
0.0074
0.1287
0.8026
0.0038
0.0018
car
0.0664
0.0076
0.0025
0.0093
0.9015
0.0127
clutter
0.4081
0.2373
0.1134
0.0090
0.0026
0.2296
Precision/Correctness
0.924
0.963
0.826
0.942
0.961
0.717
Recall/Completeness
0.955
0.988
0.912
0.803
0.901
0.230
F1
0.939
0.976
0.867
0.867
0.930
0.348

Overall accuracy 0.921

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9301
0.0139
0.0233
0.0260
0.0019
0.0049
building
0.0132
0.9772
0.0045
0.0029
0.0000
0.0021
low_veg
0.0793
0.0128
0.8205
0.0821
0.0001
0.0052
tree
0.0542
0.0063
0.1141
0.8232
0.0016
0.0006
car
0.1641
0.0005
0.0008
0.0135
0.8151
0.0061
clutter
0.2899
0.2613
0.1672
0.0129
0.0270
0.2417
Precision/Correctness
0.888
0.945
0.846
0.797
0.924
0.692
Recall/Completeness
0.930
0.977
0.821
0.823
0.815
0.242
F1
0.908
0.961
0.833
0.810
0.866
0.358

Overall accuracy 0.884

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9479
0.0087
0.0192
0.0199
0.0003
0.0040
building
0.0085
0.9845
0.0030
0.0021
0.0000
0.0019
low_veg
0.0719
0.0078
0.8434
0.0726
0.0000
0.0043
tree
0.0386
0.0045
0.1016
0.8536
0.0013
0.0004
car
0.0575
0.0003
0.0001
0.0102
0.9251
0.0068
clutter
0.2530
0.2566
0.1597
0.0100
0.0334
0.2873
Precision/Correctness
0.917
0.963
0.872
0.827
0.948
0.713
Recall/Completeness
0.948
0.985
0.843
0.854
0.925
0.287
F1
0.932
0.973
0.858
0.840
0.936
0.410

Overall accuracy 0.910

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9368
0.0075
0.0196
0.0251
0.0014
0.0096
building
0.0172
0.9761
0.0038
0.0020
0.0001
0.0009
low_veg
0.0922
0.0089
0.8124
0.0771
0.0001
0.0092
tree
0.0529
0.0052
0.0859
0.8486
0.0021
0.0053
car
0.1658
0.0131
0.0125
0.0258
0.7722
0.0105
clutter
0.2400
0.2021
0.1141
0.0453
0.0022
0.3963
Precision/Correctness
0.824
0.924
0.849
0.846
0.940
0.834
Recall/Completeness
0.937
0.976
0.812
0.849
0.772
0.396
F1
0.877
0.950
0.830
0.847
0.848
0.537

Overall accuracy 0.862

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_14_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9539
0.0043
0.0136
0.0191
0.0002
0.0089
building
0.0147
0.9811
0.0021
0.0014
0.0000
0.0007
low_veg
0.0751
0.0044
0.8412
0.0711
0.0000
0.0081
tree
0.0416
0.0042
0.0765
0.8715
0.0015
0.0047
car
0.0599
0.0143
0.0116
0.0194
0.8829
0.0119
clutter
0.2245
0.1731
0.1033
0.0437
0.0022
0.4532
Precision/Correctness
0.856
0.946
0.873
0.870
0.967
0.854
Recall/Completeness
0.954
0.981
0.841
0.872
0.883
0.453
F1
0.902
0.963
0.857
0.871
0.923
0.592

Overall accuracy 0.888

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_15_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9283
0.0113
0.0379
0.0175
0.0021
0.0027
building
0.0138
0.9720
0.0063
0.0051
0.0000
0.0027
low_veg
0.0571
0.0070
0.8072
0.1051
0.0001
0.0234
tree
0.0419
0.0050
0.0787
0.8714
0.0010
0.0020
car
0.1650
0.0120
0.0038
0.0193
0.7908
0.0092
clutter
0.1859
0.1451
0.0928
0.0142
0.0039
0.5580
Precision/Correctness
0.916
0.935
0.781
0.803
0.942
0.875
Recall/Completeness
0.928
0.972
0.807
0.871
0.791
0.558
F1
0.922
0.953
0.794
0.836
0.860
0.681

Overall accuracy 0.886

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_5_15_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9438
0.0070
0.0334
0.0132
0.0006
0.0020
building
0.0088
0.9805
0.0046
0.0038
0.0000
0.0022
low_veg
0.0378
0.0022
0.8390
0.0979
0.0000
0.0231
tree
0.0310
0.0030
0.0687
0.8950
0.0007
0.0016
car
0.0600
0.0135
0.0009
0.0160
0.8989
0.0107
clutter
0.1685
0.1110
0.0787
0.0116
0.0036
0.6266
Precision/Correctness
0.942
0.958
0.806
0.833
0.971
0.892
Recall/Completeness
0.944
0.981
0.839
0.895
0.899
0.627
F1
0.943
0.969
0.822
0.863
0.934
0.736

Overall accuracy 0.911

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9452
0.0159
0.0180
0.0152
0.0012
0.0046
building
0.0312
0.9492
0.0155
0.0026
0.0000
0.0015
low_veg
0.0256
0.0054
0.9358
0.0321
0.0000
0.0011
tree
0.0917
0.0028
0.0810
0.8235
0.0007
0.0004
car
0.1532
0.0113
0.0002
0.0110
0.8162
0.0080
clutter
0.3582
0.0562
0.3066
0.0318
0.0049
0.2423
Precision/Correctness
0.914
0.962
0.893
0.869
0.941
0.667
Recall/Completeness
0.945
0.949
0.936
0.823
0.816
0.242
F1
0.929
0.956
0.914
0.846
0.874
0.356

Overall accuracy 0.914

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9594
0.0130
0.0134
0.0101
0.0003
0.0039
building
0.0266
0.9577
0.0124
0.0021
0.0000
0.0012
low_veg
0.0169
0.0024
0.9537
0.0262
0.0000
0.0008
tree
0.0712
0.0016
0.0694
0.8570
0.0005
0.0003
car
0.0551
0.0142
0.0000
0.0074
0.9140
0.0092
clutter
0.3413
0.0547
0.3125
0.0294
0.0042
0.2578
Precision/Correctness
0.934
0.972
0.911
0.899
0.975
0.694
Recall/Completeness
0.959
0.958
0.954
0.857
0.914
0.258
F1
0.946
0.965
0.932
0.878
0.944
0.376

Overall accuracy 0.933

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9593
0.0096
0.0128
0.0106
0.0039
0.0037
building
0.0345
0.9580
0.0043
0.0015
0.0004
0.0012
low_veg
0.0778
0.0114
0.8426
0.0605
0.0005
0.0071
tree
0.0511
0.0085
0.0822
0.8548
0.0034
0.0000
car
0.1290
0.0023
0.0022
0.0162
0.8459
0.0045
clutter
0.3512
0.2587
0.1299
0.0209
0.0117
0.2276
Precision/Correctness
0.905
0.959
0.883
0.845
0.920
0.633
Recall/Completeness
0.959
0.958
0.843
0.855
0.846
0.228
F1
0.931
0.958
0.862
0.850
0.881
0.335

Overall accuracy 0.910

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_14_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9737
0.0060
0.0088
0.0076
0.0007
0.0033
building
0.0298
0.9655
0.0026
0.0010
0.0002
0.0010
low_veg
0.0593
0.0049
0.8762
0.0533
0.0002
0.0062
tree
0.0388
0.0070
0.0699
0.8818
0.0024
0.0000
car
0.0349
0.0016
0.0009
0.0125
0.9450
0.0052
clutter
0.3368
0.2847
0.1068
0.0167
0.0123
0.2428
Precision/Correctness
0.928
0.969
0.909
0.875
0.966
0.654
Recall/Completeness
0.974
0.965
0.876
0.882
0.945
0.243
F1
0.950
0.967
0.892
0.878
0.956
0.354

Overall accuracy 0.932

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_15_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9321
0.0077
0.0216
0.0095
0.0041
0.0249
building
0.0206
0.9212
0.0496
0.0031
0.0000
0.0055
low_veg
0.0755
0.0040
0.8369
0.0512
0.0005
0.0319
tree
0.0555
0.0056
0.1360
0.7969
0.0032
0.0027
car
0.1285
0.0033
0.0052
0.0208
0.8330
0.0091
clutter
0.1483
0.0247
0.1427
0.0056
0.0042
0.6744
Precision/Correctness
0.904
0.973
0.715
0.906
0.914
0.715
Recall/Completeness
0.932
0.921
0.837
0.797
0.833
0.674
F1
0.918
0.946
0.771
0.848
0.871
0.694

Overall accuracy 0.876

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_6_15_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9478
0.0053
0.0161
0.0060
0.0008
0.0240
building
0.0150
0.9300
0.0482
0.0022
0.0000
0.0046
low_veg
0.0566
0.0014
0.8696
0.0412
0.0000
0.0311
tree
0.0444
0.0041
0.1199
0.8267
0.0025
0.0024
car
0.0419
0.0034
0.0008
0.0156
0.9285
0.0098
clutter
0.1373
0.0238
0.1317
0.0040
0.0041
0.6991
Precision/Correctness
0.928
0.981
0.733
0.933
0.961
0.733
Recall/Completeness
0.948
0.930
0.870
0.827
0.929
0.699
F1
0.938
0.955
0.796
0.877
0.944
0.716

Overall accuracy 0.898

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_7_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9201
0.0334
0.0061
0.0300
0.0018
0.0087
building
0.0123
0.9577
0.0034
0.0022
0.0000
0.0244
low_veg
0.0684
0.0100
0.7376
0.1794
0.0001
0.0046
tree
0.0252
0.0052
0.0930
0.8745
0.0002
0.0019
car
0.1235
0.0242
0.0004
0.0386
0.8132
0.0000
clutter
0.2219
0.0183
0.0804
0.0027
0.0091
0.6676
Precision/Correctness
0.932
0.915
0.803
0.665
0.867
0.853
Recall/Completeness
0.920
0.958
0.738
0.874
0.813
0.668
F1
0.926
0.936
0.769
0.755
0.839
0.749

Overall accuracy 0.882

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image


=============================================================

Tile top_potsdam_7_13_class.tif, reference set: no_boundary


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9342
0.0302
0.0041
0.0225
0.0005
0.0085
building
0.0094
0.9626
0.0022
0.0014
0.0000
0.0243
low_veg
0.0541
0.0062
0.7656
0.1700
0.0000
0.0041
tree
0.0161
0.0036
0.0856
0.8925
0.0001
0.0021
car
0.0430
0.0306
0.0000
0.0302
0.8962
0.0000
clutter
0.2109
0.0171
0.0794
0.0021
0.0088
0.6817
Precision/Correctness
0.943
0.925
0.820
0.697
0.900
0.858
Recall/Completeness
0.934
0.963
0.766
0.893
0.896
0.682
F1
0.938
0.944
0.792
0.783
0.898
0.760

Overall accuracy 0.898

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image