The 3D semantic instance prediction task involves detecting and segmenting the object in an 3D scan mesh.

Evaluation and metrics

Similarly to the ScanNet benchmark in ScanNet200 our evaluation ranks all methods according to the average precision for each class. We report the mean average precision AP at overlap 0.25 (AP 25%), overlap 0.5 (AP 50%), and over overlaps in the range [0.5:0.95:0.05] (AP) for all 200 categories. Note that multiple predictions of the same ground truth instance are penalized as false positives.



This table lists the benchmark results for the ScanNet200 3D semantic instance scenario.




Method Infoavg ap 25%head ap 25%common ap 25%tail ap 25%chairtabledoorcouchcabinetshelfdeskoffice chairbedpillowsinkpicturewindowtoiletbookshelfmonitorcurtainbookarmchaircoffee tableboxrefrigeratorlampkitchen cabinettowelclothestvnightstandcounterdresserstoolcushionplantceilingbathtubend tabledining tablekeyboardbagbackpacktoilet paperprintertv standwhiteboardblanketshower curtaintrash canclosetstairsmicrowavestoveshoecomputer towerbottlebinottomanbenchboardwashing machinemirrorcopierbasketsofa chairfile cabinetfanlaptopshowerpaperpersonpaper towel dispenserovenblindsrackplateblackboardpianosuitcaserailradiatorrecycling bincontainerwardrobesoap dispensertelephonebucketclockstandlightlaundry basketpipeclothes dryerguitartoilet paper holderseatspeakercolumnbicycleladderbathroom stallshower wallcupjacketstorage bincoffee makerdishwasherpaper towel rollmachinematwindowsillbartoasterbulletin boardironing boardfireplacesoap dishkitchen counterdoorframetoilet paper dispensermini fridgefire extinguisherballhatshower curtain rodwater coolerpaper cuttertrayshower doorpillarledgetoaster ovenmousetoilet seat cover dispenserfurniturecartstorage containerscaletissue boxlight switchcratepower outletdecorationsignprojectorcloset doorvacuum cleanercandleplungerstuffed animalheadphonesdish rackbroomguitar caserange hooddustpanhair dryerwater bottlehandicap barpurseventshower floorwater pitchermailboxbowlpaper bagalarm clockmusic standprojector screendividerlaundry detergentbathroom counterobjectbathroom vanitycloset walllaundry hamperbathroom stall doorceiling lighttrash bindumbbellstair railtubebathroom cabinetcd casecloset rodcoffee kettlestructureshower headkeyboard pianocase of water bottlescoat rackstorage organizerfolded chairfire alarmpower stripcalendarposterpotted plantluggagemattress
sort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysorted bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort by
DINO3D-Scannet200copyleft0.511 30.685 30.484 10.331 30.892 20.821 20.890 30.907 50.629 10.468 20.905 10.001 71.000 10.816 10.968 20.863 30.811 40.944 80.596 90.960 50.778 30.532 20.719 90.481 10.851 80.803 10.873 10.850 10.421 10.806 50.856 60.111 60.761 20.677 10.000 40.944 10.861 41.000 10.220 20.708 40.856 50.220 10.864 10.579 11.000 10.764 100.655 40.327 41.000 10.911 30.244 10.667 90.923 10.857 10.702 10.889 30.496 20.048 20.355 100.494 20.794 40.798 31.000 10.042 40.264 70.817 60.683 20.675 10.167 40.000 50.700 10.824 30.417 60.000 50.000 40.764 10.000 70.500 20.699 30.789 60.079 80.472 10.845 40.930 10.000 30.667 10.000 50.412 20.000 20.163 51.000 10.000 50.419 10.500 21.000 10.777 20.576 30.867 40.378 20.334 40.028 30.764 40.542 10.559 10.000 40.800 10.528 40.000 30.346 50.714 10.125 40.756 40.754 50.866 40.750 10.600 30.500 10.500 11.000 10.667 11.000 10.000 10.298 10.000 50.250 50.194 20.000 70.850 30.000 50.250 50.595 10.000 30.063 10.860 40.000 10.714 20.000 10.944 10.750 10.000 10.974 10.000 10.000 10.857 40.655 30.719 90.250 30.014 60.000 11.000 10.000 10.142 40.744 20.200 70.746 30.436 30.221 60.798 10.500 70.011 40.000 10.385 70.000 10.000 20.000 50.792 60.663 10.000 50.000 20.200 50.000 40.000 41.000 10.000 2
Jinyuan Qu, Hongyang Li, Xingyu Chen, Shilong Liu, Yukai Shi, Tianhe Ren, Ruitao Jing and Lei Zhang: SegDINO3D: 3D Instance Segmentation Empowered by Both Image-Level and Object-Level 2D Features. AAAI 2026
ODIN - Ins200permissive0.451 50.637 60.407 40.277 50.742 100.699 70.855 50.826 100.626 20.441 30.742 70.003 60.941 70.637 50.910 60.616 90.679 70.944 80.695 60.877 70.763 40.357 60.723 80.475 20.779 90.494 50.782 60.795 40.334 20.824 30.867 50.108 70.701 40.638 20.000 40.873 20.749 60.667 100.203 30.500 50.886 40.116 30.583 90.571 20.688 51.000 10.760 10.162 71.000 10.852 50.078 70.833 50.887 20.778 20.577 50.859 80.550 10.000 70.542 40.028 90.667 70.874 11.000 10.125 20.232 80.870 30.406 60.337 70.167 40.000 50.671 30.742 60.500 30.000 50.000 40.528 21.000 10.417 80.597 40.872 20.275 20.000 80.800 60.850 20.000 30.528 20.000 50.215 70.000 20.238 40.667 40.000 50.019 60.250 71.000 10.429 70.599 20.778 50.221 50.370 30.284 10.278 100.400 70.125 40.000 40.200 70.404 60.000 30.250 70.714 10.500 10.504 70.769 40.677 70.750 10.963 10.500 10.000 20.500 90.333 91.000 10.000 10.000 80.438 10.500 10.000 41.000 10.333 70.226 20.250 50.250 50.000 30.000 30.668 70.000 10.494 90.000 10.000 70.750 10.000 10.833 60.000 10.000 10.777 70.333 50.944 30.000 50.333 20.000 11.000 10.000 10.089 70.407 80.600 10.823 20.080 60.264 50.469 60.717 20.000 60.000 10.500 50.000 10.000 20.000 51.000 10.125 20.333 10.000 20.200 50.000 40.000 41.000 10.000 2
CompetitorFormer-2000.469 40.676 40.401 50.296 40.901 10.729 60.885 40.829 80.380 70.320 50.873 30.400 10.998 30.711 30.980 10.847 40.854 11.000 10.696 50.989 20.759 50.556 10.806 40.240 50.918 40.650 40.818 50.629 50.224 50.839 20.933 10.247 30.711 30.540 40.021 30.543 90.900 30.903 90.118 50.125 60.916 10.057 70.692 40.410 80.747 41.000 10.664 30.424 10.933 80.839 60.207 30.703 80.748 70.700 80.610 30.869 50.270 60.068 10.878 10.244 60.794 40.698 51.000 10.000 50.325 50.770 90.482 40.452 40.025 100.015 40.293 50.829 20.663 21.000 10.013 10.385 30.250 60.500 20.491 70.850 40.214 70.131 50.878 10.617 50.000 30.085 60.009 40.278 60.000 20.295 21.000 10.000 50.160 30.500 20.500 40.342 80.534 40.901 20.474 10.222 60.011 50.724 50.542 10.125 40.083 30.336 60.500 50.083 20.565 20.587 30.500 10.827 20.829 20.750 50.508 50.018 70.500 10.000 21.000 10.667 11.000 10.000 10.173 20.286 20.500 10.000 40.125 60.489 50.000 50.500 10.269 40.000 30.050 20.834 50.000 10.581 60.000 10.677 20.467 90.000 10.886 50.000 10.000 10.820 60.144 81.000 11.000 10.103 40.000 11.000 10.000 10.175 20.410 70.330 60.701 50.257 40.292 40.285 90.574 60.157 10.000 10.863 10.000 10.056 10.250 41.000 10.000 40.109 20.000 20.400 20.025 30.000 41.000 10.000 2
ACGP-ScanNet2000.544 10.737 10.483 20.381 10.801 80.859 10.921 20.912 40.536 50.483 10.846 60.036 30.996 40.699 40.955 30.929 10.842 21.000 10.834 10.993 10.858 10.517 30.838 10.396 30.968 20.682 30.860 20.840 20.292 30.800 60.825 70.213 40.573 50.552 30.000 40.738 40.918 21.000 10.064 81.000 10.897 30.184 20.747 30.424 60.835 31.000 10.718 20.388 31.000 10.915 20.140 60.721 70.851 40.778 20.666 20.902 20.405 40.035 40.511 50.307 50.903 20.680 61.000 10.708 10.403 20.931 20.658 30.510 31.000 10.089 10.671 30.765 51.000 10.528 30.006 30.204 41.000 10.500 20.759 10.854 30.590 10.461 20.850 30.767 30.042 10.086 50.000 50.462 10.000 20.349 11.000 10.007 30.341 20.444 61.000 10.759 30.371 90.867 30.367 30.462 10.000 70.903 20.443 60.500 30.250 20.600 20.809 10.500 10.944 10.540 40.000 50.944 10.905 10.944 20.677 30.637 20.500 10.000 21.000 10.507 31.000 10.000 10.140 30.000 50.500 10.000 41.000 11.000 10.143 40.146 70.396 20.000 30.000 31.000 10.000 10.782 10.000 10.638 40.677 50.000 10.974 10.000 10.000 10.959 30.903 10.884 51.000 10.472 10.000 10.250 60.000 10.185 10.718 30.391 50.604 60.189 50.206 70.500 50.637 30.064 20.000 10.667 30.000 10.000 21.000 11.000 10.050 30.000 50.000 20.317 40.144 20.024 31.000 10.008 1
Mask3D Scannet2000.445 60.653 50.392 60.254 60.844 40.746 50.818 60.888 70.556 40.262 60.890 20.025 41.000 10.608 60.930 40.694 70.721 50.930 100.686 70.966 40.615 90.440 50.725 70.201 60.890 60.414 90.827 40.552 60.158 100.806 40.924 20.042 80.512 70.412 100.226 10.604 70.830 51.000 10.125 40.792 20.815 60.097 40.648 50.551 50.354 91.000 10.630 50.241 61.000 10.853 40.204 40.974 40.841 50.778 20.358 70.927 10.300 50.045 30.640 20.363 30.745 60.710 41.000 10.000 50.330 40.943 10.315 70.600 21.000 10.027 30.080 100.556 100.500 30.409 40.000 40.194 51.000 10.500 20.493 60.761 70.053 90.042 70.780 70.454 60.009 20.333 30.050 10.321 40.000 20.084 60.552 70.008 20.027 50.750 10.500 40.442 60.657 10.765 60.120 70.183 80.021 41.000 10.510 50.016 60.000 40.400 30.619 30.000 30.396 40.290 60.000 50.741 50.699 61.000 10.260 60.017 80.125 100.000 20.792 80.399 81.000 10.000 10.049 60.265 30.063 80.000 41.000 10.335 60.381 10.500 10.250 50.004 20.000 30.727 60.000 10.538 70.000 10.188 50.677 50.000 10.930 40.000 10.000 10.966 20.391 40.908 40.000 50.028 50.000 11.000 10.000 10.152 30.451 50.458 30.971 10.573 20.606 10.167 100.625 40.004 50.000 10.058 100.000 10.000 21.000 11.000 10.000 40.056 30.000 20.200 50.309 10.000 41.000 10.000 2
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
Volt-SPFormerpermissive0.527 20.731 20.475 30.342 20.826 50.803 30.942 10.950 20.594 30.321 40.867 40.008 50.994 50.767 20.926 50.874 20.815 31.000 10.810 20.973 30.856 20.510 40.825 20.346 40.923 30.799 20.843 30.812 30.262 40.923 10.921 30.279 20.901 10.500 60.000 40.801 30.937 11.000 10.329 10.000 70.903 20.076 50.789 20.565 30.907 21.000 10.614 60.413 21.000 10.937 10.214 20.629 100.878 30.725 70.579 40.880 40.433 30.020 60.400 60.547 11.000 10.843 21.000 10.125 20.343 30.855 40.750 10.449 51.000 10.057 20.700 10.802 40.500 30.850 20.011 20.047 61.000 11.000 10.715 20.875 10.255 30.099 60.857 20.738 40.000 30.056 70.025 20.372 30.250 10.279 30.667 40.002 40.000 70.250 70.500 41.000 10.391 80.737 70.309 40.397 20.000 70.817 30.542 10.557 21.000 10.400 30.681 20.000 30.500 30.519 50.500 10.773 30.818 30.884 30.656 40.510 40.500 10.000 21.000 10.472 41.000 10.000 10.027 70.000 50.331 40.000 41.000 11.000 10.000 50.500 10.304 30.000 30.000 31.000 10.000 10.714 20.000 10.677 20.750 10.000 10.944 30.000 10.000 11.000 10.764 20.833 60.250 30.278 30.000 11.000 10.000 10.103 50.753 10.600 10.508 90.638 10.167 80.458 70.741 10.019 30.000 10.850 20.000 10.000 21.000 11.000 10.000 40.028 40.000 20.200 50.000 40.250 11.000 10.000 2
Kadir Yilmaz, Adrian Kruse, Tristan Höfer, Daan de Geus, Bastian Leibe: Volume Transformer: Revisiting Vanilla Transformers for 3D Scene Understanding.
TD3D Scannet200permissive0.379 70.603 70.306 70.190 70.885 30.755 40.800 70.958 10.390 60.260 70.866 50.232 20.979 60.523 80.869 80.559 100.689 61.000 10.795 30.905 60.748 60.173 100.825 30.173 70.970 10.457 60.615 70.456 70.200 60.621 90.906 40.553 10.517 60.510 50.220 20.715 50.706 71.000 10.113 60.792 20.717 70.073 60.635 60.557 40.638 61.000 10.205 100.146 81.000 10.769 100.186 51.000 10.710 100.778 20.415 60.834 90.226 70.021 50.590 30.356 40.817 30.477 101.000 10.000 50.635 10.843 50.427 50.270 90.125 60.000 50.102 81.000 10.125 70.000 50.000 40.000 70.000 70.125 90.370 80.622 100.221 40.196 40.836 50.288 70.000 30.093 40.020 30.294 50.000 20.075 70.667 40.038 10.111 40.250 70.000 90.526 50.495 60.908 10.111 80.259 50.003 60.667 60.045 100.000 70.000 40.400 30.274 80.000 30.274 60.226 70.000 50.520 60.302 100.731 60.103 80.458 50.500 10.000 21.000 10.472 40.792 80.000 10.088 50.061 40.250 50.009 30.250 50.333 70.181 30.396 40.051 70.012 10.000 30.458 90.000 10.424 100.000 10.101 60.390 100.000 10.833 60.000 10.000 10.857 40.222 71.000 10.000 50.003 70.000 10.000 70.000 10.102 60.275 100.400 40.735 40.061 80.433 30.533 40.625 40.000 60.000 10.259 90.000 10.000 20.000 50.500 70.000 40.000 51.000 10.600 10.000 40.250 10.000 70.000 2
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
Minkowski 34D Inst.permissive0.280 90.488 90.192 100.124 90.804 70.518 90.772 100.904 60.337 100.191 90.443 90.000 80.861 90.502 90.868 90.669 80.587 90.997 60.467 100.828 100.732 70.342 80.745 60.119 100.918 40.404 100.419 90.398 80.172 80.618 100.743 90.167 50.077 100.500 60.000 40.568 80.506 101.000 10.044 90.000 70.502 90.010 90.593 80.284 100.305 100.903 90.213 90.142 90.981 70.790 90.000 91.000 10.715 90.538 100.346 90.830 100.067 80.000 70.400 60.074 80.333 90.551 71.000 10.000 50.292 60.777 80.118 100.317 80.100 80.000 50.191 70.648 80.000 80.000 50.000 40.000 70.000 70.500 20.213 100.825 50.021 100.333 30.648 100.098 90.000 30.000 80.000 50.077 80.000 20.000 100.150 100.000 50.000 70.000 100.225 70.281 90.447 70.000 100.090 90.148 90.000 70.479 90.542 10.000 70.000 40.200 70.131 100.000 30.250 70.000 90.000 50.159 100.396 90.677 70.021 90.000 90.500 10.000 21.000 10.442 70.125 100.000 10.000 80.000 50.000 90.333 10.000 70.528 40.000 50.000 80.000 80.000 30.000 30.200 100.000 10.516 80.000 10.000 70.500 70.000 10.833 60.000 10.000 10.286 90.083 90.750 70.000 50.000 80.000 10.000 70.000 10.059 100.445 60.200 70.535 80.070 70.167 80.385 80.375 80.000 60.000 10.333 80.000 10.000 20.000 50.500 70.000 40.000 50.000 20.200 50.000 40.000 40.000 70.000 2
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
CSC-Pretrain Inst.permissive0.275 100.466 100.218 90.110 100.783 90.383 100.783 90.829 90.367 90.168 100.305 100.000 80.661 100.413 100.869 70.719 50.546 100.997 60.685 80.841 90.555 100.277 90.768 50.132 80.779 90.448 80.364 100.212 100.161 90.768 70.692 100.000 90.395 80.500 60.000 40.450 100.591 81.000 10.020 100.000 70.423 100.007 100.625 70.420 70.505 81.000 10.353 70.119 100.571 90.819 70.014 81.000 10.774 60.689 90.311 100.866 60.067 80.000 70.400 60.000 100.278 100.501 81.000 10.000 50.162 100.584 100.286 80.206 100.125 60.000 50.084 90.649 70.000 80.000 50.000 40.000 70.000 70.125 90.312 90.727 80.221 50.000 80.667 90.114 80.000 30.000 80.000 50.065 100.000 20.004 90.278 80.000 50.000 70.500 20.000 90.571 40.000 100.250 90.019 100.145 100.000 70.667 60.200 90.000 70.000 40.200 70.258 90.000 30.000 90.000 90.000 50.369 90.429 80.613 90.000 100.000 90.500 10.000 20.500 90.333 90.500 90.000 10.106 40.000 50.000 90.000 40.000 70.333 70.000 50.000 80.000 80.000 30.000 30.918 30.000 10.638 40.000 10.000 70.750 10.000 10.833 60.000 10.000 10.143 100.000 100.750 70.000 50.000 80.000 10.000 70.000 10.063 90.377 90.200 70.222 100.055 90.500 20.677 30.250 90.000 60.000 10.500 50.000 10.000 20.000 50.500 70.000 40.000 50.000 20.115 100.000 40.000 40.000 70.000 2
Ji Hou, Benjamin Graham, Matthias Nießner, Saining Xie: Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts. CVPR 2021
LGround Inst.permissive0.314 80.529 80.225 80.155 80.810 60.625 80.798 80.940 30.372 80.217 80.484 80.000 80.927 80.528 70.826 100.694 60.605 81.000 10.731 40.846 80.716 80.350 70.589 100.123 90.857 70.457 70.578 80.376 90.183 70.765 80.800 80.000 90.278 90.500 60.000 40.659 60.569 91.000 10.093 70.000 70.539 80.010 80.578 100.378 90.571 71.000 10.337 80.252 50.530 100.814 80.000 90.744 60.743 80.746 60.346 80.863 70.067 80.000 70.400 60.167 70.667 70.488 91.000 10.000 50.208 90.783 70.166 90.375 60.071 90.000 50.200 60.607 90.000 80.000 50.000 40.000 71.000 10.500 20.517 50.716 90.221 50.000 80.706 80.085 100.000 30.000 80.000 50.077 90.000 20.063 80.278 80.000 50.000 70.500 20.083 80.181 100.515 50.286 80.144 60.219 70.042 20.582 80.400 70.000 70.000 40.000 100.305 70.000 30.000 90.036 80.000 50.413 80.500 70.533 100.250 70.200 60.500 10.000 21.000 10.472 41.000 10.000 10.000 80.000 50.250 50.000 40.000 70.333 70.000 50.000 80.000 80.000 30.000 30.600 80.000 10.594 50.000 10.000 70.500 70.000 10.647 100.000 10.000 10.429 80.333 50.500 100.000 50.000 80.000 10.000 70.000 10.069 80.696 40.050 100.556 70.031 100.042 100.750 20.250 90.000 60.000 10.630 40.000 10.000 20.000 50.500 70.000 40.000 50.000 20.400 20.000 40.000 40.000 70.000 2
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild.