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%alarm clockarmchairbackpackbagballbarbasketbathroom cabinetbathroom counterbathroom stallbathroom stall doorbathroom vanitybathtubbedbenchbicyclebinblackboardblanketblindsboardbookbookshelfbottlebowlboxbroombucketbulletin boardcabinetcalendarcandlecartcase of water bottlescd caseceilingceiling lightchairclockclosetcloset doorcloset rodcloset wallclothesclothes dryercoat rackcoffee kettlecoffee makercoffee tablecolumncomputer towercontainercopiercouchcountercratecupcurtaincushiondecorationdeskdining tabledish rackdishwasherdividerdoordoorframedresserdumbbelldustpanend tablefanfile cabinetfire alarmfire extinguisherfireplacefolded chairfurnitureguitarguitar casehair dryerhandicap barhatheadphonesironing boardjacketkeyboardkeyboard pianokitchen cabinetkitchen counterladderlamplaptoplaundry basketlaundry detergentlaundry hamperledgelightlight switchluggagemachinemailboxmatmattressmicrowavemini fridgemirrormonitormousemusic standnightstandobjectoffice chairottomanovenpaperpaper bagpaper cutterpaper towel dispenserpaper towel rollpersonpianopicturepillarpillowpipeplantplateplungerposterpotted plantpower outletpower stripprinterprojectorprojector screenpurserackradiatorrailrange hoodrecycling binrefrigeratorscaleseatshelfshoeshowershower curtainshower curtain rodshower doorshower floorshower headshower wallsignsinksoap dishsoap dispensersofa chairspeakerstair railstairsstandstoolstorage binstorage containerstorage organizerstovestructurestuffed animalsuitcasetabletelephonetissue boxtoastertoaster oventoilettoilet papertoilet paper dispensertoilet paper holdertoilet seat cover dispensertoweltrash bintrash cantraytubetvtv standvacuum cleanerventwardrobewashing machinewater bottlewater coolerwater pitcherwhiteboardwindowwindowsill
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 bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort bysort 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 by
AQ3D-ScanNet2000.491 20.675 10.458 10.308 30.823 20.153 30.500 10.500 10.833 10.000 10.104 50.444 70.834 11.000 10.955 40.564 41.000 10.025 50.533 50.371 40.333 50.458 30.459 10.754 30.258 60.167 50.307 30.000 10.083 10.605 30.596 10.250 70.000 10.500 10.000 30.832 40.073 60.907 10.009 30.005 70.553 10.310 10.667 20.056 20.000 10.489 90.804 50.250 60.754 111.000 10.860 40.049 60.362 20.824 10.160 30.026 80.796 10.023 70.750 10.600 10.846 30.530 30.778 20.438 60.077 70.473 50.991 10.141 30.318 60.200 40.000 20.517 40.500 10.319 30.829 30.917 60.648 20.506 50.737 30.698 30.434 60.000 20.000 10.517 30.105 30.431 10.000 50.563 20.479 30.400 30.831 50.658 40.521 30.919 50.000 40.000 10.896 20.074 70.328 21.000 10.071 20.667 11.000 10.868 20.083 50.428 51.000 10.925 10.768 20.147 10.879 20.000 70.640 20.000 20.141 80.000 40.667 60.000 31.000 10.133 10.765 31.000 10.534 20.929 21.000 10.029 20.305 40.551 31.000 11.000 11.000 10.714 71.000 10.713 10.000 30.662 60.818 50.347 30.172 30.281 21.000 10.000 50.576 30.004 20.000 30.000 20.778 20.000 20.000 10.671 30.712 20.698 31.000 10.716 11.000 10.517 50.750 30.737 30.433 30.934 10.000 10.083 20.764 51.000 11.000 10.000 10.664 11.000 10.000 10.333 80.667 30.498 60.666 20.923 1
ACGP-ScanNet2000.494 10.656 20.453 30.345 10.723 40.178 20.500 10.500 10.708 20.000 10.020 90.592 30.695 41.000 10.933 60.511 61.000 10.035 41.000 10.378 30.528 30.307 60.442 20.834 10.260 51.000 10.313 20.000 10.042 20.540 50.484 61.000 10.000 10.396 40.000 30.853 10.189 50.771 70.010 20.023 30.471 20.276 30.817 10.000 50.000 10.903 20.838 10.444 50.817 41.000 10.831 80.212 10.310 40.718 20.000 50.396 20.687 51.000 10.677 60.400 20.864 20.545 20.573 60.500 40.062 80.511 30.931 20.144 20.637 20.317 30.000 20.638 30.000 60.396 10.819 51.000 10.655 10.532 30.714 40.680 40.510 30.000 20.000 10.387 60.071 40.405 20.036 40.250 30.449 50.600 20.836 40.677 30.566 20.963 10.000 40.000 10.743 80.148 10.035 41.000 10.089 10.472 21.000 10.765 50.500 20.671 10.500 30.871 20.633 50.040 70.736 40.204 40.627 30.008 10.146 70.024 30.835 20.000 30.250 70.005 20.842 20.974 20.590 10.893 51.000 10.007 40.344 10.495 51.000 11.000 10.792 70.835 51.000 10.643 20.000 30.823 10.844 40.360 20.341 20.088 90.703 60.000 50.548 40.000 30.000 30.000 20.667 60.000 20.000 10.702 20.746 10.664 41.000 10.304 50.997 50.424 70.750 30.840 10.204 80.914 30.000 10.064 30.800 21.000 11.000 10.000 10.461 20.903 30.000 10.504 30.850 10.618 20.633 30.657 2
Rongkun Yang, Ye Zhang, Longguang Wang, Zhiheng Fu, Lian Xu, Yulan Guo: Beyond Context Bias: Adaptive Instance Placement for Robust 3D Instance Segmentation.
DINO3D-Scannet200copyleft0.454 40.587 40.453 20.296 40.851 10.200 10.500 10.000 40.042 50.000 10.378 10.545 40.729 31.000 10.981 30.355 111.000 10.046 20.000 80.248 50.000 60.494 20.381 50.586 100.496 20.250 30.409 10.000 10.000 40.714 10.572 21.000 10.000 10.250 60.050 20.793 50.436 30.871 30.000 50.216 10.284 30.290 20.083 80.000 50.000 10.764 40.716 100.500 20.842 31.000 10.891 20.096 30.361 30.690 50.000 50.595 10.753 30.708 40.750 10.400 20.845 50.475 50.728 30.750 10.214 20.683 20.743 60.000 50.400 50.200 40.500 10.944 10.125 50.327 20.823 40.792 70.602 40.662 10.777 20.803 10.675 10.000 20.000 10.200 80.298 10.324 40.000 50.000 50.000 100.800 10.824 60.750 10.507 40.937 30.000 40.000 10.779 70.116 30.001 80.417 70.000 60.014 71.000 10.816 30.548 10.600 20.500 30.771 50.773 10.117 20.944 10.764 10.571 50.000 20.250 50.000 41.000 10.063 11.000 10.000 30.720 50.974 20.079 90.918 40.000 80.000 60.312 30.616 10.125 51.000 11.000 10.857 20.000 60.594 40.000 30.767 30.845 30.264 70.419 10.177 60.667 80.000 50.677 10.000 30.194 10.000 20.857 10.000 20.000 10.563 50.703 30.835 20.850 40.346 40.944 90.499 60.866 20.777 20.221 70.911 40.000 10.011 50.721 60.764 100.520 80.000 10.442 30.405 90.000 10.667 10.655 40.473 70.614 50.437 6
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
TD3D Scannet200permissive0.320 80.501 80.264 80.164 80.506 80.062 70.500 10.000 40.000 60.000 10.208 30.431 80.252 91.000 10.733 90.587 30.000 80.008 70.000 80.106 60.000 60.356 40.123 100.686 40.101 80.000 60.152 80.000 10.000 40.226 70.280 90.000 80.000 10.250 60.000 30.619 70.061 90.841 40.000 50.000 80.167 70.194 60.333 60.000 50.000 10.667 60.820 30.250 60.790 81.000 10.879 30.077 50.094 90.708 30.217 20.049 70.634 60.792 20.331 100.033 110.716 80.159 80.396 80.331 80.099 60.415 60.842 30.000 50.458 40.542 10.000 20.101 70.000 60.218 60.513 80.500 80.458 80.104 80.516 50.456 60.268 100.000 20.000 10.400 40.022 60.233 60.143 30.000 50.677 10.400 30.504 110.095 90.083 110.890 70.061 30.000 10.906 10.076 60.231 30.125 80.000 60.003 80.792 90.881 10.000 80.098 90.125 100.498 100.459 80.063 40.715 60.000 70.241 90.000 20.396 40.063 20.605 70.000 30.000 80.000 30.448 110.629 90.202 70.967 10.250 60.038 10.192 70.185 80.083 81.000 11.000 10.857 20.000 60.470 60.012 10.565 90.798 70.621 10.111 50.500 11.000 10.017 30.509 60.000 30.008 21.000 10.525 80.000 20.000 10.332 90.679 40.264 80.333 70.267 61.000 10.549 30.299 110.387 80.328 40.744 100.000 10.000 70.435 111.000 10.283 100.000 10.196 40.817 40.000 10.472 40.222 80.123 100.560 70.156 8
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
CompetitorFormer-2000.415 50.574 50.370 60.274 50.632 50.054 80.500 10.083 30.000 60.000 10.260 20.541 50.410 60.903 80.987 20.885 10.500 50.064 10.250 60.378 21.000 10.243 70.428 30.599 80.218 71.000 10.196 60.000 10.000 40.587 40.318 81.000 10.000 10.500 10.000 30.845 20.232 40.885 20.005 40.015 40.216 60.183 70.399 50.082 10.000 10.724 50.806 40.500 20.869 21.000 10.779 90.095 40.443 10.685 60.021 40.269 40.704 40.083 60.467 90.400 20.846 40.551 10.663 40.261 90.103 50.482 40.758 50.025 40.018 80.400 20.000 20.677 20.500 10.207 70.881 21.000 10.600 50.648 20.144 100.641 50.452 40.000 20.000 10.327 70.142 20.209 70.000 50.083 40.215 80.317 70.748 70.508 60.484 50.957 20.000 40.000 10.833 40.132 20.400 10.663 30.015 50.103 51.000 10.759 60.125 40.286 60.500 30.830 40.651 40.089 30.540 90.380 30.581 40.000 20.500 10.000 40.745 30.050 21.000 10.000 30.622 80.694 70.213 60.870 70.125 70.000 60.205 60.562 20.000 90.933 91.000 10.820 60.250 50.347 80.000 30.731 50.877 10.289 50.160 40.186 50.684 70.008 40.538 50.000 30.000 30.000 20.700 50.056 10.000 10.491 60.584 80.602 60.489 50.565 21.000 10.311 90.750 30.583 60.292 50.832 70.000 10.157 10.780 41.000 10.625 60.000 10.131 50.794 50.000 10.667 10.071 100.545 40.682 10.462 4
Volt-SPFormerpermissive0.475 30.630 30.451 40.314 20.772 30.068 60.500 10.000 40.125 30.000 10.107 40.524 60.742 21.000 10.994 10.400 70.500 50.019 61.000 10.410 10.667 20.500 10.423 40.811 20.412 30.250 30.281 40.000 10.000 40.519 60.541 41.000 10.000 10.331 50.000 30.841 30.638 10.806 60.000 50.014 50.241 50.245 40.333 60.028 40.000 10.817 30.825 20.250 60.799 71.000 10.847 50.129 20.294 50.702 40.000 50.304 30.755 20.000 80.750 10.400 20.923 10.482 40.900 10.208 100.319 10.750 10.823 40.000 50.510 30.200 40.000 20.500 50.500 10.300 50.903 11.000 10.564 60.372 61.000 10.787 20.449 50.250 10.000 10.600 10.026 50.375 30.000 51.000 10.455 40.400 30.878 20.641 50.612 10.894 60.000 40.000 10.800 50.078 50.008 60.500 40.056 30.278 41.000 10.797 40.500 20.585 31.000 10.869 30.735 30.056 50.768 30.043 60.714 10.000 20.500 10.250 10.683 50.000 31.000 10.000 30.853 10.944 40.255 40.923 31.000 10.002 50.224 50.499 40.250 41.000 11.000 10.857 21.000 10.613 30.000 30.818 20.857 20.343 40.000 70.209 40.629 100.025 20.500 70.000 30.000 30.000 20.725 40.000 20.000 10.716 10.666 60.651 51.000 10.500 30.990 80.565 20.750 30.699 40.167 90.930 20.000 10.019 40.784 31.000 11.000 10.000 10.099 61.000 10.000 10.472 40.764 20.546 30.621 40.452 5
Kadir Yilmaz, Adrian Kruse, Tristan Höfer, Daan de Geus, Bastian Leibe: Volume Transformer: Revisiting Vanilla Transformers for 3D Scene Understanding.
Mask3D Scannet2000.388 60.542 60.357 70.237 70.610 60.091 50.125 110.000 40.000 60.000 10.065 70.668 10.451 51.000 10.955 40.640 20.500 50.039 30.125 70.063 80.409 40.311 50.291 60.609 70.266 40.000 60.163 70.000 10.008 30.044 80.496 51.000 10.000 10.018 80.000 30.756 60.573 20.808 50.000 50.010 60.042 90.130 90.552 40.042 30.000 11.000 10.725 80.750 10.883 11.000 10.832 70.024 80.107 70.614 80.226 10.250 50.628 70.792 20.677 60.400 20.741 70.278 70.511 70.077 110.111 40.313 80.715 70.302 10.017 90.200 40.000 20.188 60.000 60.178 80.736 71.000 10.615 30.514 40.409 60.380 110.600 20.000 20.000 10.400 40.013 70.254 50.381 10.000 50.123 90.400 30.839 30.258 70.463 60.926 40.265 20.000 10.857 30.099 40.021 50.500 40.027 40.028 61.000 10.502 110.016 70.076 100.500 30.612 60.578 60.005 80.597 70.194 50.497 60.000 20.500 10.000 40.323 100.000 31.000 10.000 30.748 40.708 60.050 100.890 61.000 10.008 30.151 90.301 71.000 11.000 10.792 70.945 11.000 10.511 50.004 20.753 40.776 80.287 60.020 60.003 100.974 40.033 10.412 110.000 30.000 30.000 20.667 60.000 20.000 10.491 70.676 50.352 70.335 60.060 80.822 110.527 41.000 10.517 70.606 10.853 60.000 10.004 60.806 11.000 10.727 40.000 10.042 70.739 60.000 10.399 70.391 50.504 50.591 60.571 3
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
ODIN - Ins200permissive0.381 70.507 70.375 50.237 60.484 90.108 40.500 10.000 40.125 30.000 10.058 80.647 20.385 70.667 90.853 70.542 51.000 10.000 81.000 10.093 70.000 60.028 100.274 70.682 50.550 10.000 60.269 50.000 10.000 40.714 10.566 31.000 10.000 10.500 10.125 10.585 80.066 80.653 110.083 10.049 20.264 40.227 50.667 20.000 50.000 10.278 110.723 90.250 60.786 91.000 10.744 110.039 70.209 60.494 100.000 50.250 50.446 80.500 50.750 10.200 80.780 60.333 60.602 50.469 50.163 30.406 70.530 90.000 50.668 10.200 40.000 20.000 80.500 10.313 40.769 61.000 10.511 70.196 70.286 70.393 100.337 70.000 20.000 10.600 10.000 80.174 80.226 20.000 50.579 20.200 80.887 10.750 10.428 70.782 80.438 10.000 10.795 60.063 80.003 70.500 40.000 60.333 31.000 10.742 70.083 50.585 30.417 90.448 110.496 70.055 60.734 50.472 20.174 100.000 20.250 50.000 40.688 40.000 31.000 10.000 30.631 70.667 80.275 30.694 111.000 10.000 60.328 20.422 60.000 91.000 10.500 90.638 80.000 60.391 70.000 30.582 80.800 60.208 100.000 70.246 30.667 80.000 50.638 20.167 10.000 30.000 20.778 20.000 20.000 10.563 40.614 70.841 10.333 70.250 70.938 100.569 10.500 90.695 50.264 60.863 50.000 10.000 70.550 101.000 10.668 50.000 10.000 80.667 70.000 10.333 80.333 60.665 10.434 80.264 7
Minkowski 34D Inst.permissive0.203 110.369 100.134 110.078 110.479 100.003 100.500 10.000 40.000 60.000 10.100 60.371 90.300 80.667 90.746 80.400 70.000 80.000 80.000 80.031 90.000 60.074 90.165 90.413 110.000 100.000 60.070 100.000 10.000 40.000 90.221 110.000 80.000 10.000 90.000 30.372 110.070 70.706 90.000 50.000 80.000 110.123 100.033 110.000 50.000 10.422 100.732 70.000 100.778 101.000 10.845 60.000 90.090 100.636 70.000 50.000 90.158 100.000 80.250 110.050 100.693 90.123 100.051 110.385 70.009 100.118 110.406 110.000 50.000 100.200 40.000 20.000 80.000 60.133 100.307 110.500 80.251 100.000 100.281 80.402 90.317 80.000 20.000 10.000 90.000 80.060 100.000 50.000 50.396 60.200 80.669 80.021 100.218 100.720 110.000 40.000 10.696 90.025 100.000 90.000 90.000 60.000 90.125 110.596 80.000 80.191 70.500 30.595 70.369 100.000 90.500 100.000 70.143 110.000 20.000 90.000 40.226 110.000 30.000 80.000 30.701 60.511 100.000 110.851 90.000 80.000 60.150 100.052 110.100 70.981 80.500 90.286 90.000 60.000 110.000 30.545 100.522 110.250 80.000 70.000 110.522 110.000 50.500 70.000 30.000 30.000 20.282 110.000 20.000 10.178 110.382 100.018 110.056 100.000 90.997 50.107 110.677 70.313 100.000 100.726 110.000 10.000 70.583 90.903 90.200 110.000 10.000 80.333 100.000 10.442 60.083 90.109 110.387 100.000 11
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
CSC-Pretrain Inst.permissive0.209 100.361 110.157 100.085 100.506 70.007 90.500 10.000 40.000 60.000 10.000 110.093 110.221 100.667 90.524 110.400 70.000 80.000 80.000 80.004 100.000 60.000 110.109 110.589 90.000 100.000 60.059 110.000 10.000 40.000 90.322 70.000 80.000 10.000 90.000 30.405 90.055 100.700 100.000 50.000 80.028 100.091 110.083 80.000 50.000 10.667 60.768 60.000 100.807 61.000 10.776 100.000 90.000 110.340 110.000 50.000 90.103 110.000 80.750 10.200 80.634 110.053 110.246 90.677 30.006 110.198 90.432 100.000 50.000 100.050 100.000 20.000 80.000 60.111 110.356 100.500 80.188 110.000 100.220 90.448 70.050 110.000 20.000 10.000 90.000 80.032 110.000 50.000 50.396 60.000 100.573 100.000 110.228 90.747 100.000 40.000 10.573 110.021 110.000 90.000 90.000 60.000 90.500 100.573 90.000 80.000 110.125 100.592 80.364 110.000 90.450 110.000 70.364 70.000 20.000 90.000 40.340 90.000 30.000 80.000 30.610 90.833 50.221 50.702 100.000 80.000 60.135 110.094 100.125 50.571 100.500 90.143 110.000 60.125 90.000 30.618 70.667 100.115 110.000 70.125 71.000 10.000 50.500 70.000 30.000 30.000 20.502 100.000 20.000 10.312 100.248 110.050 100.000 110.000 90.997 50.420 80.500 90.149 110.451 20.748 80.000 10.000 70.636 80.667 110.600 70.000 10.000 80.278 110.000 10.333 80.000 110.294 80.381 110.110 9
Ji Hou, Benjamin Graham, Matthias Nießner, Saining Xie: Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts. CVPR 2021
LGround Inst.permissive0.246 90.413 90.170 90.130 90.455 110.003 110.500 10.000 40.000 60.000 10.017 100.333 100.111 111.000 10.681 100.400 70.000 80.000 81.000 10.003 110.000 60.167 80.190 80.637 60.067 90.000 60.081 90.000 10.000 40.000 90.264 100.000 80.000 10.000 90.000 30.387 100.031 110.754 80.000 50.000 80.151 80.135 80.056 100.000 50.000 10.582 80.589 110.500 20.815 51.000 10.903 10.000 90.097 80.588 90.000 50.000 90.234 90.000 80.500 80.400 20.682 100.156 90.159 100.750 10.046 90.125 100.660 80.000 50.200 70.000 110.000 20.000 80.000 60.164 90.402 90.500 80.373 90.025 90.143 110.426 80.317 80.000 20.000 10.000 90.000 80.063 90.000 50.000 50.000 100.000 100.575 90.250 80.241 80.772 90.000 40.000 10.653 100.034 90.000 90.000 90.000 60.000 91.000 10.561 100.000 80.100 80.500 30.541 90.452 90.000 90.581 80.000 70.364 70.000 20.000 90.000 40.571 80.000 30.000 80.000 30.568 100.511 100.167 80.857 80.000 80.000 60.164 80.112 90.000 90.530 111.000 10.286 90.000 60.125 90.000 30.464 110.706 90.208 90.000 70.125 70.744 50.000 50.500 70.000 30.000 30.000 20.511 90.000 20.000 10.344 80.541 90.068 90.333 70.000 91.000 10.196 100.533 80.318 90.000 100.748 90.000 10.000 70.690 71.000 10.400 90.000 10.000 80.667 70.000 10.333 80.333 60.270 90.399 90.083 10
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild.