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 50%head ap 50%common ap 50%tail ap 50%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 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 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
ACGP-ScanNet2000.494 10.656 10.453 20.345 10.771 60.746 10.864 20.831 70.484 50.344 10.687 40.035 30.933 50.633 40.823 10.871 10.633 20.997 40.834 10.963 10.718 10.442 10.838 10.313 20.893 40.680 30.655 10.840 10.276 20.800 20.743 70.212 10.573 50.548 30.000 40.736 30.853 11.000 10.062 71.000 10.819 40.178 20.723 30.424 60.835 21.000 10.618 20.378 31.000 10.914 20.023 30.703 50.836 40.667 50.495 40.817 40.260 50.035 40.511 50.307 50.903 20.566 21.000 10.708 10.360 20.931 10.511 30.510 31.000 10.089 10.671 10.765 41.000 10.528 30.005 10.204 41.000 10.500 20.702 20.842 20.590 10.461 10.844 40.664 30.042 10.010 20.000 50.405 10.000 20.040 60.817 10.007 30.341 20.444 51.000 10.714 30.020 80.643 10.310 30.396 10.000 20.903 20.400 10.500 20.250 20.600 20.657 10.500 10.304 40.540 40.000 50.532 30.545 20.750 30.677 30.637 20.500 10.000 20.792 60.504 31.000 10.000 10.071 30.000 40.396 30.000 31.000 11.000 10.036 40.146 70.396 20.000 30.000 31.000 10.000 10.627 20.000 10.638 30.677 50.000 10.974 10.000 10.000 10.835 50.850 10.449 41.000 10.472 10.000 10.250 60.000 10.148 10.695 30.387 50.592 30.189 50.204 70.500 40.088 80.064 20.000 10.471 10.000 10.000 21.000 11.000 10.000 30.000 40.000 20.317 30.144 20.024 31.000 10.008 1
Mask3D Scannet2000.388 50.542 50.357 60.237 60.808 40.676 40.741 60.832 60.496 40.151 80.628 60.021 40.955 40.578 50.753 40.612 50.591 50.822 100.609 60.926 40.614 70.291 50.725 70.163 60.890 50.380 100.615 20.517 60.130 80.806 10.857 20.024 70.511 60.412 100.226 10.597 60.756 51.000 10.111 40.792 20.736 60.091 40.610 50.527 40.323 91.000 10.504 50.063 71.000 10.853 50.010 60.974 30.839 30.667 50.301 60.883 10.266 40.039 30.640 20.311 40.739 50.463 51.000 10.000 50.287 50.715 60.313 70.600 21.000 10.027 30.076 90.502 100.500 30.409 40.000 20.194 50.125 60.500 20.491 60.748 30.050 90.042 60.776 70.352 60.008 20.000 40.033 10.254 40.000 20.005 70.552 30.008 20.020 50.750 10.500 40.409 50.065 60.511 40.107 60.178 70.000 21.000 10.400 10.016 60.000 40.400 30.571 20.000 30.060 70.044 70.000 50.514 40.278 61.000 10.258 60.017 80.125 100.000 20.792 60.399 71.000 10.000 10.013 60.265 20.018 70.000 31.000 10.335 50.381 10.500 10.250 50.004 20.000 30.727 30.000 10.497 50.000 10.188 50.677 50.000 10.708 50.000 10.000 10.945 10.391 40.123 80.000 50.028 50.000 11.000 10.000 10.099 40.451 40.400 30.668 10.573 20.606 10.077 100.003 90.004 50.000 10.042 80.000 10.000 21.000 11.000 10.000 30.042 20.000 20.200 40.302 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
DINO3D-Scannet200copyleft0.454 30.587 30.453 10.296 30.871 20.703 20.845 40.891 20.572 10.312 30.753 20.001 70.981 30.773 10.767 30.771 40.614 40.944 80.586 90.937 30.690 40.381 40.716 90.409 10.918 30.803 10.602 30.777 20.290 10.721 50.779 60.096 30.728 20.677 10.000 40.944 10.793 41.000 10.214 20.708 40.823 30.200 10.851 10.499 51.000 10.764 90.473 60.248 41.000 10.911 30.216 10.667 70.824 50.857 10.616 10.842 30.496 20.046 20.355 100.494 20.405 80.507 31.000 10.042 40.264 60.743 50.683 20.675 10.125 40.000 50.600 20.816 20.417 60.000 50.000 20.764 10.000 70.500 20.563 40.720 40.079 80.442 20.845 30.835 20.000 30.000 40.000 50.324 30.000 20.117 10.083 70.000 50.419 10.500 21.000 10.777 20.378 10.594 30.361 20.327 20.000 20.764 40.400 10.548 10.000 40.800 10.437 50.000 30.346 30.714 10.125 40.662 10.475 40.866 20.750 10.400 50.500 10.500 11.000 10.667 11.000 10.000 10.298 10.000 40.250 50.194 10.000 70.850 30.000 50.250 50.595 10.000 30.063 10.520 70.000 10.571 40.000 10.944 10.750 10.000 10.974 10.000 10.000 10.857 20.655 30.000 90.250 30.014 60.000 11.000 10.000 10.116 30.729 20.200 70.545 40.436 30.221 60.750 10.177 50.011 40.000 10.284 20.000 10.000 20.000 50.792 60.050 20.000 40.000 20.200 40.000 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
CompetitorFormer-2000.415 40.574 40.370 50.274 40.885 10.584 70.846 30.779 80.318 70.205 50.704 30.400 10.987 20.651 30.731 50.830 30.682 11.000 10.599 70.957 20.685 50.428 20.806 40.196 50.870 60.641 40.600 40.583 50.183 60.780 40.833 30.095 40.663 30.538 40.021 30.540 80.845 20.903 70.103 50.083 60.881 20.054 70.632 40.311 80.745 31.000 10.545 40.378 20.933 80.832 60.015 40.684 60.748 60.700 40.562 20.869 20.218 60.064 10.885 10.243 60.794 40.484 41.000 10.000 50.289 40.758 40.482 40.452 40.000 80.015 40.286 50.759 50.663 21.000 10.000 20.380 30.250 50.500 20.491 50.622 70.213 50.131 40.877 10.602 50.000 30.005 30.008 40.209 60.000 20.089 20.399 40.000 50.160 30.500 20.500 40.144 90.260 20.347 70.443 10.207 60.000 20.724 50.400 10.125 40.083 30.317 60.462 30.083 20.565 10.587 30.500 10.648 20.551 10.750 30.508 50.018 70.500 10.000 21.000 10.667 11.000 10.000 10.142 20.000 40.500 10.000 30.125 60.489 40.000 50.500 10.269 40.000 30.050 20.625 50.000 10.581 30.000 10.677 20.467 80.000 10.694 60.000 10.000 10.820 60.071 90.215 71.000 10.103 40.000 11.000 10.000 10.132 20.410 50.327 60.541 50.232 40.292 40.261 80.186 40.157 10.000 10.216 50.000 10.056 10.250 41.000 10.000 30.082 10.000 20.400 20.025 30.000 41.000 10.000 2
Volt-SPFormerpermissive0.475 20.630 20.451 30.314 20.806 50.666 50.923 10.847 40.541 30.224 40.755 10.008 50.994 10.735 20.818 20.869 20.621 30.990 70.811 20.894 50.702 30.423 30.825 20.281 30.923 20.787 20.564 50.699 30.245 30.784 30.800 40.129 20.900 10.500 60.000 40.768 20.841 31.000 10.319 10.000 70.903 10.068 50.772 20.565 20.683 51.000 10.546 30.410 11.000 10.930 10.014 50.629 90.878 20.725 30.499 30.799 70.412 30.019 50.400 60.500 11.000 10.612 11.000 10.125 20.343 30.823 30.750 10.449 50.250 30.056 20.585 30.797 30.500 30.667 20.000 20.043 61.000 11.000 10.716 10.853 10.255 30.099 50.857 20.651 40.000 30.000 40.025 20.375 20.250 10.056 40.333 50.002 40.000 60.250 60.500 41.000 10.107 40.613 20.294 40.300 40.000 20.817 30.400 10.500 21.000 10.400 30.452 40.000 30.500 20.519 50.500 10.372 50.482 30.750 30.641 40.510 30.500 10.000 21.000 10.472 41.000 10.000 10.026 40.000 40.331 40.000 31.000 11.000 10.000 50.500 10.304 30.000 30.000 31.000 10.000 10.714 10.000 10.500 40.750 10.000 10.944 30.000 10.000 10.857 20.764 20.455 30.250 30.278 30.000 11.000 10.000 10.078 50.742 10.600 10.524 60.638 10.167 80.208 90.209 30.019 30.000 10.241 40.000 10.000 21.000 11.000 10.000 30.028 30.000 20.200 40.000 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.
ODIN - Ins200permissive0.381 60.507 60.375 40.237 50.653 100.614 60.780 50.744 100.566 20.328 20.446 70.003 60.853 60.496 60.582 70.448 100.434 70.938 90.682 40.782 70.494 90.274 60.723 80.269 40.694 100.393 90.511 60.695 40.227 40.550 90.795 50.039 60.602 40.638 20.000 40.734 40.585 70.667 80.163 30.500 50.769 50.108 30.484 80.569 10.688 41.000 10.665 10.093 61.000 10.863 40.049 20.667 70.887 10.778 20.422 50.786 90.550 10.000 70.542 40.028 90.667 60.428 61.000 10.125 20.208 90.530 80.406 60.337 60.000 80.000 50.585 30.742 60.500 30.000 50.000 20.472 21.000 10.417 80.563 30.631 60.275 20.000 70.800 50.841 10.000 30.083 10.000 50.174 70.000 20.055 50.667 20.000 50.000 60.250 61.000 10.286 60.058 70.391 60.209 50.313 30.167 10.278 100.200 70.083 50.000 40.200 70.264 60.000 30.250 60.714 10.500 10.196 60.333 50.500 80.750 10.668 10.500 10.000 20.500 80.333 81.000 10.000 10.000 70.438 10.500 10.000 31.000 10.333 60.226 20.250 50.250 50.000 30.000 30.668 40.000 10.174 90.000 10.000 70.750 10.000 10.667 70.000 10.000 10.638 70.333 50.579 20.000 50.333 20.000 11.000 10.000 10.063 70.385 60.600 10.647 20.066 70.264 50.469 50.246 20.000 60.000 10.264 30.000 10.000 20.000 51.000 10.125 10.000 40.000 20.200 40.000 40.000 41.000 10.000 2
TD3D Scannet200permissive0.320 70.501 70.264 70.164 70.841 30.679 30.716 70.879 30.280 80.192 60.634 50.231 20.733 80.459 70.565 80.498 90.560 61.000 10.686 30.890 60.708 20.123 90.820 30.152 70.967 10.456 50.458 70.387 70.194 50.435 100.906 10.077 50.396 70.509 50.217 20.715 50.619 61.000 10.099 60.792 20.513 70.062 60.506 70.549 30.605 61.000 10.123 90.106 51.000 10.744 90.000 71.000 10.504 100.525 70.185 70.790 80.101 70.008 60.587 30.356 30.817 30.083 101.000 10.000 50.621 10.842 20.415 50.268 90.083 70.000 50.098 80.881 10.125 70.000 50.000 20.000 70.000 70.125 90.332 80.448 100.202 60.196 30.798 60.264 70.000 30.000 40.017 30.233 50.000 20.063 30.333 50.038 10.111 40.250 60.000 70.516 40.208 30.470 50.094 80.218 50.000 20.667 60.033 100.000 70.000 40.400 30.156 70.000 30.267 50.226 60.000 50.104 70.159 70.299 100.095 80.458 40.500 10.000 21.000 10.472 40.792 80.000 10.022 50.061 30.250 50.008 20.250 50.333 60.143 30.396 40.049 70.012 10.000 30.283 90.000 10.241 80.000 10.101 60.331 90.000 10.629 80.000 10.000 10.857 20.222 70.677 10.000 50.003 70.000 10.000 70.000 10.076 60.252 80.400 30.431 70.061 80.328 30.331 70.500 10.000 60.000 10.167 60.000 10.000 20.000 50.500 70.000 30.000 41.000 10.542 10.000 40.063 20.000 70.000 2
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
LGround Inst.permissive0.246 80.413 80.170 80.130 80.754 70.541 80.682 90.903 10.264 90.164 70.234 80.000 80.681 90.452 80.464 100.541 80.399 81.000 10.637 50.772 80.588 80.190 70.589 100.081 80.857 70.426 70.373 80.318 80.135 70.690 60.653 90.000 80.159 90.500 60.000 40.581 70.387 91.000 10.046 80.000 70.402 80.003 100.455 100.196 90.571 71.000 10.270 80.003 100.530 100.748 80.000 70.744 40.575 80.511 80.112 80.815 50.067 80.000 70.400 60.167 70.667 60.241 71.000 10.000 50.208 80.660 70.125 90.317 70.000 80.000 50.100 70.561 90.000 80.000 50.000 20.000 71.000 10.500 20.344 70.568 90.167 70.000 70.706 80.068 80.000 30.000 40.000 50.063 80.000 20.000 80.056 90.000 50.000 60.500 20.000 70.143 100.017 90.125 80.097 70.164 80.000 20.582 80.400 10.000 70.000 40.000 90.083 90.000 30.000 80.000 80.000 50.025 80.156 80.533 70.250 70.200 60.500 10.000 21.000 10.333 81.000 10.000 10.000 70.000 40.000 80.000 30.000 70.333 60.000 50.000 80.000 80.000 30.000 30.400 80.000 10.364 60.000 10.000 70.500 70.000 10.511 90.000 10.000 10.286 80.333 50.000 90.000 50.000 80.000 10.000 70.000 10.034 80.111 100.000 80.333 90.031 100.000 90.750 10.125 60.000 60.000 10.151 70.000 10.000 20.000 50.500 70.000 30.000 40.000 20.000 100.000 40.000 40.000 70.000 2
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild.
Minkowski 34D Inst.permissive0.203 100.369 90.134 100.078 100.706 80.382 90.693 80.845 50.221 100.150 90.158 90.000 80.746 70.369 90.545 90.595 60.387 90.997 40.413 100.720 100.636 60.165 80.732 60.070 90.851 80.402 80.251 90.313 90.123 90.583 80.696 80.000 80.051 100.500 60.000 40.500 90.372 100.667 80.009 90.000 70.307 100.003 90.479 90.107 100.226 100.903 80.109 100.031 80.981 70.726 100.000 70.522 100.669 70.282 100.052 100.778 100.000 90.000 70.400 60.074 80.333 90.218 91.000 10.000 50.250 70.406 100.118 100.317 70.100 60.000 50.191 60.596 70.000 80.000 50.000 20.000 70.000 70.500 20.178 100.701 50.000 100.000 70.522 100.018 100.000 30.000 40.000 50.060 90.000 20.000 80.033 100.000 50.000 60.000 90.000 70.281 70.100 50.000 100.090 90.133 90.000 20.422 90.050 90.000 70.000 40.200 70.000 100.000 30.000 80.000 80.000 50.000 90.123 90.677 60.021 90.000 90.500 10.000 20.500 80.442 60.125 100.000 10.000 70.000 40.000 80.000 30.000 70.056 90.000 50.000 80.000 80.000 30.000 30.200 100.000 10.143 100.000 10.000 70.250 100.000 10.511 90.000 10.000 10.286 80.083 80.396 50.000 50.000 80.000 10.000 70.000 10.025 90.300 70.000 80.371 80.070 60.000 90.385 60.000 100.000 60.000 10.000 100.000 10.000 20.000 50.500 70.000 30.000 40.000 20.200 40.000 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.permissive0.209 90.361 100.157 90.085 90.700 90.248 100.634 100.776 90.322 60.135 100.103 100.000 80.524 100.364 100.618 60.592 70.381 100.997 40.589 80.747 90.340 100.109 100.768 50.059 100.702 90.448 60.188 100.149 100.091 100.636 70.573 100.000 80.246 80.500 60.000 40.450 100.405 80.667 80.006 100.000 70.356 90.007 80.506 60.420 70.340 80.667 100.294 70.004 90.571 90.748 70.000 71.000 10.573 90.502 90.094 90.807 60.000 90.000 70.400 60.000 100.278 100.228 81.000 10.000 50.115 100.432 90.198 80.050 100.125 40.000 50.000 100.573 80.000 80.000 50.000 20.000 70.000 70.125 90.312 90.610 80.221 40.000 70.667 90.050 90.000 30.000 40.000 50.032 100.000 20.000 80.083 70.000 50.000 60.000 90.000 70.220 80.000 100.125 80.000 100.111 100.000 20.667 60.200 70.000 70.000 40.000 90.110 80.000 30.000 80.000 80.000 50.000 90.053 100.500 80.000 100.000 90.500 10.000 20.500 80.333 80.500 90.000 10.000 70.000 40.000 80.000 30.000 70.000 100.000 50.000 80.000 80.000 30.000 30.600 60.000 10.364 60.000 10.000 70.750 10.000 10.833 40.000 10.000 10.143 100.000 100.396 50.000 50.000 80.000 10.000 70.000 10.021 100.221 90.000 80.093 100.055 90.451 20.677 30.125 60.000 60.000 10.028 90.000 10.000 20.000 50.500 70.000 30.000 40.000 20.050 90.000 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