Detailled semantic labeling (2D) result


Name K. Nogueira et al.
Affiliation Universidade Federal de Minas Gerais, Brazil
Abbreviation UFMG_2
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
0.895
0.019
0.032
0.022
0.003
0.030
building
0.020
0.949
0.007
0.011
0.002
0.011
low_veg
0.053
0.013
0.815
0.097
0.000
0.022
tree
0.042
0.007
0.147
0.793
0.004
0.008
car
0.077
0.006
0.002
0.008
0.897
0.010
clutter
0.333
0.102
0.106
0.032
0.029
0.399
Precision/Correctness
0.876
0.940
0.804
0.810
0.834
0.505
Recall/Completeness
0.895
0.949
0.815
0.793
0.897
0.399
F1
0.885
0.945
0.809
0.801
0.864
0.446

Overall accuracy 0.851

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
0.911
0.015
0.027
0.017
0.001
0.029
building
0.017
0.957
0.005
0.010
0.002
0.010
low_veg
0.041
0.008
0.840
0.090
0.000
0.021
tree
0.033
0.005
0.135
0.817
0.003
0.007
car
0.014
0.006
0.000
0.005
0.963
0.011
clutter
0.336
0.086
0.092
0.028
0.028
0.431
Precision/Correctness
0.898
0.956
0.826
0.835
0.858
0.513
Recall/Completeness
0.911
0.957
0.840
0.817
0.963
0.431
F1
0.905
0.956
0.833
0.826
0.908
0.468

Overall accuracy 0.872
=============================================================

Tile top_potsdam_2_13_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9021
0.0224
0.0293
0.0146
0.0028
0.0288
building
0.0139
0.9498
0.0058
0.0181
0.0016
0.0108
low_veg
0.0499
0.0160
0.8290
0.0686
0.0004
0.0361
tree
0.0444
0.0140
0.1662
0.7593
0.0024
0.0137
car
0.1002
0.0078
0.0010
0.0037
0.8524
0.0348
clutter
0.1570
0.1605
0.1108
0.0207
0.0611
0.4900
Precision/Correctness
0.873
0.912
0.837
0.862
0.777
0.285
Recall/Completeness
0.902
0.950
0.829
0.759
0.852
0.490
F1
0.887
0.931
0.833
0.808
0.813
0.361

Overall accuracy 0.847

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
0.9249
0.0173
0.0213
0.0102
0.0008
0.0255
building
0.0105
0.9588
0.0037
0.0169
0.0014
0.0087
low_veg
0.0417
0.0115
0.8529
0.0583
0.0003
0.0352
tree
0.0359
0.0123
0.1487
0.7883
0.0019
0.0129
car
0.0440
0.0077
0.0001
0.0016
0.9079
0.0387
clutter
0.1194
0.1513
0.0815
0.0165
0.0685
0.5628
Precision/Correctness
0.900
0.935
0.861
0.886
0.815
0.268
Recall/Completeness
0.925
0.959
0.853
0.788
0.908
0.563
F1
0.912
0.947
0.857
0.834
0.859
0.363

Overall accuracy 0.872

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
0.8707
0.0093
0.0625
0.0151
0.0017
0.0407
building
0.0083
0.9665
0.0071
0.0097
0.0003
0.0080
low_veg
0.0418
0.0049
0.8775
0.0485
0.0002
0.0272
tree
0.0261
0.0059
0.2296
0.7288
0.0013
0.0083
car
0.0747
0.0109
0.0016
0.0036
0.9020
0.0072
clutter
0.1388
0.2545
0.1256
0.0406
0.0843
0.3562
Precision/Correctness
0.827
0.911
0.797
0.918
0.830
0.057
Recall/Completeness
0.871
0.967
0.877
0.729
0.902
0.356
F1
0.848
0.938
0.835
0.812
0.864
0.099

Overall accuracy 0.828

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
0.8929
0.0062
0.0492
0.0096
0.0005
0.0416
building
0.0055
0.9778
0.0039
0.0073
0.0001
0.0054
low_veg
0.0361
0.0028
0.8922
0.0413
0.0001
0.0275
tree
0.0221
0.0052
0.2155
0.7482
0.0010
0.0080
car
0.0192
0.0120
0.0002
0.0018
0.9602
0.0067
clutter
0.1035
0.2849
0.1003
0.0440
0.0983
0.3690
Precision/Correctness
0.849
0.935
0.812
0.934
0.860
0.043
Recall/Completeness
0.893
0.978
0.892
0.748
0.960
0.369
F1
0.870
0.956
0.850
0.831
0.907
0.077

Overall accuracy 0.845

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
0.8814
0.0210
0.0388
0.0256
0.0037
0.0296
building
0.0153
0.9636
0.0053
0.0061
0.0007
0.0089
low_veg
0.0626
0.0248
0.7844
0.1082
0.0003
0.0197
tree
0.0279
0.0074
0.1116
0.8407
0.0045
0.0080
car
0.0727
0.0061
0.0011
0.0101
0.9016
0.0085
clutter
0.3173
0.1579
0.1954
0.0353
0.0236
0.2705
Precision/Correctness
0.869
0.935
0.814
0.851
0.849
0.284
Recall/Completeness
0.881
0.964
0.784
0.841
0.902
0.271
F1
0.875
0.949
0.799
0.846
0.874
0.277

Overall accuracy 0.856

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
0.9081
0.0144
0.0297
0.0181
0.0006
0.0292
building
0.0127
0.9703
0.0037
0.0048
0.0006
0.0079
low_veg
0.0511
0.0146
0.8175
0.0984
0.0001
0.0182
tree
0.0214
0.0057
0.0975
0.8641
0.0038
0.0075
car
0.0144
0.0065
0.0003
0.0059
0.9649
0.0079
clutter
0.3273
0.1487
0.1696
0.0285
0.0259
0.2999
Precision/Correctness
0.897
0.958
0.842
0.877
0.886
0.265
Recall/Completeness
0.908
0.970
0.817
0.864
0.965
0.300
F1
0.902
0.964
0.830
0.871
0.924
0.282

Overall accuracy 0.883

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
0.8667
0.0278
0.0469
0.0298
0.0049
0.0240
building
0.0137
0.9693
0.0043
0.0067
0.0005
0.0055
low_veg
0.0367
0.0152
0.8481
0.0877
0.0002
0.0120
tree
0.0279
0.0086
0.2457
0.7044
0.0030
0.0105
car
0.0637
0.0108
0.0023
0.0130
0.9023
0.0080
clutter
0.2330
0.1547
0.1449
0.0365
0.0228
0.4082
Precision/Correctness
0.867
0.926
0.745
0.847
0.802
0.501
Recall/Completeness
0.867
0.969
0.848
0.704
0.902
0.408
F1
0.867
0.947
0.793
0.769
0.849
0.450

Overall accuracy 0.829

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
0.8937
0.0215
0.0370
0.0226
0.0018
0.0233
building
0.0114
0.9760
0.0022
0.0056
0.0004
0.0043
low_veg
0.0260
0.0095
0.8758
0.0781
0.0001
0.0105
tree
0.0232
0.0074
0.2349
0.7222
0.0027
0.0098
car
0.0101
0.0110
0.0006
0.0072
0.9642
0.0069
clutter
0.2247
0.1432
0.1171
0.0298
0.0238
0.4614
Precision/Correctness
0.896
0.947
0.762
0.873
0.835
0.511
Recall/Completeness
0.894
0.976
0.876
0.722
0.964
0.461
F1
0.895
0.961
0.815
0.790
0.895
0.485

Overall accuracy 0.852

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
0.9118
0.0145
0.0301
0.0303
0.0016
0.0117
building
0.0139
0.9629
0.0053
0.0099
0.0008
0.0071
low_veg
0.0476
0.0197
0.7681
0.1563
0.0003
0.0079
tree
0.0384
0.0032
0.0557
0.8911
0.0040
0.0076
car
0.0789
0.0041
0.0021
0.0134
0.8937
0.0078
clutter
0.4108
0.1127
0.1461
0.0664
0.0226
0.2414
Precision/Correctness
0.857
0.954
0.799
0.741
0.872
0.665
Recall/Completeness
0.912
0.963
0.768
0.891
0.894
0.241
F1
0.884
0.959
0.783
0.809
0.883
0.354

Overall accuracy 0.859

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
0.9292
0.0110
0.0245
0.0237
0.0003
0.0112
building
0.0091
0.9727
0.0025
0.0085
0.0007
0.0065
low_veg
0.0362
0.0145
0.7948
0.1478
0.0001
0.0066
tree
0.0283
0.0019
0.0445
0.9159
0.0030
0.0065
car
0.0102
0.0038
0.0003
0.0080
0.9694
0.0083
clutter
0.4170
0.0969
0.1355
0.0592
0.0233
0.2681
Precision/Correctness
0.883
0.967
0.832
0.772
0.890
0.684
Recall/Completeness
0.929
0.973
0.795
0.916
0.969
0.268
F1
0.906
0.970
0.813
0.838
0.928
0.385

Overall accuracy 0.885

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
0.8756
0.0163
0.0553
0.0187
0.0029
0.0312
building
0.0165
0.9614
0.0029
0.0069
0.0054
0.0069
low_veg
0.0904
0.0151
0.7543
0.0938
0.0009
0.0456
tree
0.0422
0.0051
0.1305
0.8095
0.0044
0.0082
car
0.0779
0.0077
0.0012
0.0120
0.8894
0.0119
clutter
0.5772
0.0388
0.0760
0.0167
0.0052
0.2861
Precision/Correctness
0.710
0.939
0.788
0.837
0.857
0.567
Recall/Completeness
0.876
0.961
0.754
0.810
0.889
0.286
F1
0.784
0.950
0.771
0.823
0.873
0.380

Overall accuracy 0.794

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
0.8909
0.0114
0.0518
0.0143
0.0003
0.0313
building
0.0136
0.9688
0.0012
0.0054
0.0052
0.0057
low_veg
0.0760
0.0099
0.7814
0.0860
0.0005
0.0461
tree
0.0338
0.0038
0.1140
0.8376
0.0037
0.0072
car
0.0200
0.0082
0.0007
0.0082
0.9507
0.0123
clutter
0.5955
0.0350
0.0626
0.0128
0.0040
0.2901
Precision/Correctness
0.722
0.956
0.809
0.862
0.883
0.581
Recall/Completeness
0.891
0.969
0.781
0.838
0.951
0.290
F1
0.797
0.962
0.795
0.850
0.915
0.387

Overall accuracy 0.813

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
0.8907
0.0201
0.0502
0.0108
0.0030
0.0252
building
0.0074
0.9705
0.0053
0.0130
0.0003
0.0034
low_veg
0.0531
0.0249
0.8457
0.0618
0.0007
0.0138
tree
0.0735
0.0065
0.1456
0.7575
0.0108
0.0061
car
0.0825
0.0042
0.0023
0.0019
0.8984
0.0107
clutter
0.3629
0.2236
0.0971
0.0307
0.0252
0.2605
Precision/Correctness
0.902
0.943
0.775
0.838
0.865
0.367
Recall/Completeness
0.891
0.971
0.846
0.758
0.898
0.261
F1
0.897
0.956
0.809
0.796
0.881
0.305

Overall accuracy 0.873

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
0.9063
0.0143
0.0458
0.0077
0.0008
0.0251
building
0.0052
0.9764
0.0036
0.0120
0.0002
0.0025
low_veg
0.0335
0.0163
0.8832
0.0550
0.0002
0.0118
tree
0.0598
0.0049
0.1257
0.7949
0.0091
0.0055
car
0.0172
0.0042
0.0001
0.0007
0.9653
0.0125
clutter
0.3795
0.2256
0.0805
0.0269
0.0218
0.2656
Precision/Correctness
0.928
0.959
0.799
0.865
0.900
0.342
Recall/Completeness
0.906
0.976
0.883
0.795
0.965
0.266
F1
0.917
0.968
0.839
0.828
0.932
0.299

Overall accuracy 0.897

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
0.8953
0.0218
0.0315
0.0250
0.0022
0.0242
building
0.0163
0.9483
0.0040
0.0088
0.0037
0.0190
low_veg
0.0574
0.0129
0.8011
0.1110
0.0003
0.0172
tree
0.0547
0.0047
0.1279
0.8058
0.0040
0.0029
car
0.0948
0.0054
0.0005
0.0041
0.8862
0.0091
clutter
0.2390
0.2327
0.1104
0.0324
0.1161
0.2694
Precision/Correctness
0.902
0.938
0.832
0.750
0.789
0.333
Recall/Completeness
0.895
0.948
0.801
0.806
0.886
0.269
F1
0.898
0.943
0.816
0.777
0.835
0.298

Overall accuracy 0.861

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
0.9132
0.0168
0.0269
0.0197
0.0004
0.0231
building
0.0127
0.9545
0.0028
0.0078
0.0037
0.0185
low_veg
0.0501
0.0080
0.8225
0.1037
0.0001
0.0155
tree
0.0405
0.0032
0.1148
0.8357
0.0034
0.0023
car
0.0164
0.0062
0.0000
0.0023
0.9646
0.0104
clutter
0.2090
0.2201
0.0996
0.0318
0.1398
0.2998
Precision/Correctness
0.927
0.956
0.857
0.774
0.801
0.311
Recall/Completeness
0.913
0.955
0.823
0.836
0.965
0.300
F1
0.920
0.955
0.839
0.804
0.875
0.305

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_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.8975
0.0088
0.0278
0.0313
0.0014
0.0331
building
0.0176
0.9635
0.0036
0.0047
0.0006
0.0099
low_veg
0.0722
0.0074
0.7616
0.1322
0.0004
0.0261
tree
0.0547
0.0067
0.1057
0.8137
0.0066
0.0128
car
0.0950
0.0169
0.0040
0.0066
0.8625
0.0150
clutter
0.2235
0.1576
0.1154
0.0523
0.0193
0.4319
Precision/Correctness
0.836
0.934
0.816
0.777
0.841
0.624
Recall/Completeness
0.898
0.964
0.762
0.814
0.863
0.432
F1
0.865
0.948
0.788
0.795
0.852
0.510

Overall accuracy 0.836

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
0.9128
0.0062
0.0224
0.0256
0.0002
0.0327
building
0.0154
0.9689
0.0018
0.0040
0.0005
0.0093
low_veg
0.0570
0.0039
0.7848
0.1291
0.0001
0.0252
tree
0.0447
0.0056
0.0978
0.8348
0.0055
0.0115
car
0.0162
0.0177
0.0026
0.0034
0.9428
0.0171
clutter
0.2083
0.1295
0.1042
0.0522
0.0193
0.4865
Precision/Correctness
0.865
0.953
0.838
0.797
0.858
0.639
Recall/Completeness
0.913
0.969
0.785
0.835
0.943
0.486
F1
0.888
0.961
0.811
0.816
0.898
0.552

Overall accuracy 0.859

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
0.8874
0.0067
0.0421
0.0268
0.0019
0.0351
building
0.0141
0.9528
0.0074
0.0142
0.0020
0.0095
low_veg
0.0521
0.0083
0.7608
0.1593
0.0002
0.0193
tree
0.0385
0.0035
0.0944
0.8566
0.0044
0.0026
car
0.0972
0.0044
0.0024
0.0112
0.8752
0.0096
clutter
0.3895
0.1036
0.0995
0.0324
0.0333
0.3417
Precision/Correctness
0.887
0.953
0.748
0.715
0.848
0.549
Recall/Completeness
0.887
0.953
0.761
0.857
0.875
0.342
F1
0.887
0.953
0.754
0.780
0.861
0.421

Overall accuracy 0.843

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
0.9004
0.0035
0.0381
0.0219
0.0008
0.0353
building
0.0103
0.9613
0.0062
0.0122
0.0017
0.0083
low_veg
0.0337
0.0044
0.7873
0.1557
0.0001
0.0189
tree
0.0289
0.0019
0.0851
0.8782
0.0039
0.0021
car
0.0181
0.0047
0.0002
0.0095
0.9572
0.0103
clutter
0.4047
0.0705
0.0902
0.0304
0.0353
0.3689
Precision/Correctness
0.906
0.973
0.768
0.742
0.854
0.554
Recall/Completeness
0.900
0.961
0.787
0.878
0.957
0.369
F1
0.903
0.967
0.777
0.804
0.902
0.443

Overall accuracy 0.864

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
0.9277
0.0271
0.0143
0.0156
0.0024
0.0128
building
0.0318
0.9352
0.0082
0.0183
0.0006
0.0058
low_veg
0.0275
0.0089
0.8921
0.0701
0.0001
0.0014
tree
0.0924
0.0029
0.0873
0.8132
0.0021
0.0020
car
0.0779
0.0039
0.0002
0.0014
0.8970
0.0196
clutter
0.5091
0.0641
0.1174
0.0416
0.0321
0.2357
Precision/Correctness
0.905
0.941
0.913
0.777
0.846
0.407
Recall/Completeness
0.928
0.935
0.892
0.813
0.897
0.236
F1
0.916
0.938
0.902
0.795
0.871
0.299

Overall accuracy 0.893

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
0.9403
0.0243
0.0110
0.0108
0.0014
0.0121
building
0.0278
0.9429
0.0057
0.0179
0.0006
0.0051
low_veg
0.0185
0.0060
0.9104
0.0642
0.0000
0.0009
tree
0.0726
0.0021
0.0772
0.8446
0.0017
0.0017
car
0.0130
0.0050
0.0000
0.0005
0.9619
0.0196
clutter
0.5304
0.0578
0.0963
0.0395
0.0291
0.2470
Precision/Correctness
0.923
0.952
0.930
0.800
0.872
0.410
Recall/Completeness
0.940
0.943
0.910
0.845
0.962
0.247
F1
0.932
0.947
0.920
0.822
0.915
0.308

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_14_class.tif, reference set: full_reference


↓ predicted || reference →
imp_surf
building
low_veg
tree
car
clutter
imp_surf
0.9403
0.0053
0.0210
0.0139
0.0040
0.0154
building
0.0556
0.9169
0.0034
0.0055
0.0065
0.0121
low_veg
0.0618
0.0096
0.8113
0.1081
0.0005
0.0087
tree
0.0520
0.0073
0.0900
0.8396
0.0085
0.0026
car
0.0519
0.0026
0.0013
0.0064
0.9323
0.0055
clutter
0.3312
0.1889
0.2046
0.0457
0.0307
0.1988
Precision/Correctness
0.901
0.969
0.847
0.761
0.852
0.288
Recall/Completeness
0.940
0.917
0.811
0.840
0.932
0.199
F1
0.920
0.942
0.829
0.798
0.890
0.235

Overall accuracy 0.886

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
0.9537
0.0026
0.0170
0.0111
0.0002
0.0155
building
0.0511
0.9247
0.0015
0.0050
0.0062
0.0114
low_veg
0.0441
0.0054
0.8401
0.1024
0.0001
0.0079
tree
0.0415
0.0058
0.0794
0.8640
0.0071
0.0022
car
0.0050
0.0017
0.0005
0.0037
0.9832
0.0058
clutter
0.3192
0.2057
0.2018
0.0409
0.0291
0.2032
Precision/Correctness
0.919
0.977
0.869
0.787
0.881
0.274
Recall/Completeness
0.954
0.925
0.840
0.864
0.983
0.203
F1
0.936
0.950
0.854
0.823
0.929
0.233

Overall accuracy 0.904

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
0.8822
0.0056
0.0314
0.0215
0.0046
0.0547
building
0.0136
0.9113
0.0299
0.0301
0.0018
0.0133
low_veg
0.0435
0.0052
0.7996
0.1164
0.0012
0.0343
tree
0.0294
0.0052
0.1315
0.8251
0.0058
0.0031
car
0.0610
0.0009
0.0028
0.0112
0.9155
0.0085
clutter
0.1719
0.0192
0.1190
0.0242
0.0457
0.6200
Precision/Correctness
0.927
0.977
0.719
0.790
0.824
0.557
Recall/Completeness
0.882
0.911
0.800
0.825
0.916
0.620
F1
0.904
0.943
0.757
0.807
0.868
0.587

Overall accuracy 0.853

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
0.8977
0.0033
0.0256
0.0171
0.0005
0.0559
building
0.0095
0.9198
0.0280
0.0293
0.0016
0.0118
low_veg
0.0265
0.0033
0.8273
0.1093
0.0001
0.0334
tree
0.0212
0.0038
0.1158
0.8519
0.0047
0.0026
car
0.0090
0.0007
0.0004
0.0078
0.9730
0.0091
clutter
0.1665
0.0165
0.1121
0.0205
0.0433
0.6411
Precision/Correctness
0.946
0.985
0.738
0.814
0.861
0.565
Recall/Completeness
0.898
0.920
0.827
0.852
0.973
0.641
F1
0.921
0.952
0.780
0.833
0.914
0.601

Overall accuracy 0.873

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
0.8731
0.0430
0.0078
0.0292
0.0021
0.0449
building
0.0318
0.9266
0.0017
0.0045
0.0047
0.0308
low_veg
0.0573
0.0192
0.7098
0.2002
0.0003
0.0133
tree
0.0269
0.0068
0.0910
0.8696
0.0017
0.0041
car
0.0516
0.0066
0.0012
0.0124
0.9260
0.0023
clutter
0.2232
0.0216
0.0332
0.0020
0.0156
0.7043
Precision/Correctness
0.924
0.889
0.833
0.652
0.776
0.673
Recall/Completeness
0.873
0.927
0.710
0.870
0.926
0.704
F1
0.898
0.908
0.767
0.745
0.845
0.689

Overall accuracy 0.854

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
0.8851
0.0398
0.0059
0.0229
0.0007
0.0455
building
0.0294
0.9312
0.0011
0.0036
0.0043
0.0305
low_veg
0.0435
0.0147
0.7343
0.1941
0.0000
0.0133
tree
0.0154
0.0050
0.0823
0.8924
0.0011
0.0038
car
0.0049
0.0064
0.0004
0.0071
0.9795
0.0018
clutter
0.2156
0.0195
0.0308
0.0017
0.0141
0.7183
Precision/Correctness
0.933
0.901
0.854
0.677
0.793
0.678
Recall/Completeness
0.885
0.931
0.734
0.892
0.980
0.718
F1
0.908
0.916
0.790
0.770
0.877
0.698

Overall accuracy 0.867

Red/green image, indicating wrongly classified pixels


Original, resized true ortho image for better comparison


Classified image