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 aphead apcommon aptail apalarm 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 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 bysort bysort bysort bysort bysort bysort bysort by
CompetitorFormer-2000.328 50.439 40.303 50.223 50.543 50.044 70.333 20.044 30.000 60.000 10.099 10.444 30.296 60.850 60.722 20.820 10.444 60.047 10.083 60.183 30.562 10.243 60.312 20.380 50.192 51.000 10.143 60.000 10.000 40.484 50.259 71.000 10.000 10.500 10.000 30.650 10.221 40.771 10.004 40.010 30.043 70.120 70.366 20.054 10.000 10.689 30.641 50.500 20.663 61.000 10.673 50.049 40.400 10.479 40.014 40.267 40.455 30.083 60.400 60.400 20.663 40.243 20.464 30.192 70.076 70.427 40.620 40.025 40.013 80.322 20.000 20.677 20.333 20.178 60.808 20.556 10.356 50.345 10.119 80.346 50.312 50.000 20.000 10.305 60.116 20.137 70.000 50.065 40.171 50.314 70.575 50.487 60.303 20.820 10.000 40.000 10.655 20.088 20.373 10.430 40.011 50.103 50.835 40.569 60.125 40.123 60.500 30.774 40.504 40.019 40.465 80.353 20.475 30.000 20.500 10.000 40.712 30.050 20.667 10.000 30.396 60.555 60.120 60.786 40.069 60.000 60.182 50.390 30.000 90.831 41.000 10.679 20.111 50.110 70.000 30.450 60.868 10.277 50.083 40.069 40.471 60.001 40.428 80.000 30.000 30.000 20.421 30.043 10.000 10.358 60.456 60.518 40.237 50.256 40.945 10.271 70.632 40.534 60.208 40.730 60.000 10.140 10.658 41.000 10.452 70.000 10.082 60.441 60.000 10.472 30.060 90.454 30.469 10.384 2
DINO3D-Scannet200copyleft0.346 40.437 50.353 40.229 40.687 20.174 10.333 20.000 40.042 50.000 10.094 20.384 40.618 10.940 40.764 10.292 110.889 20.042 20.000 80.142 50.000 60.456 20.263 50.371 70.407 10.250 30.257 10.000 10.000 40.642 10.431 21.000 10.000 10.250 60.028 20.594 50.436 30.729 30.000 50.138 10.192 40.206 20.083 80.000 50.000 10.611 40.574 60.306 30.719 21.000 10.733 20.066 20.361 20.545 20.000 50.585 10.388 50.558 30.639 10.400 20.659 50.183 50.297 50.246 60.199 20.373 50.446 50.000 50.378 50.156 40.500 10.772 10.111 50.253 40.752 30.477 50.325 60.282 40.551 40.504 40.241 60.000 20.000 10.156 80.238 10.251 30.000 50.000 50.000 100.599 20.712 10.750 10.266 40.766 40.000 40.000 10.628 30.082 30.001 80.417 50.000 60.014 70.708 50.536 70.516 10.328 40.500 30.669 50.529 30.027 30.732 10.764 10.365 50.000 20.250 40.000 40.921 10.063 10.222 40.000 30.520 30.769 30.045 80.714 60.000 80.000 60.264 20.417 20.049 70.731 80.514 70.545 50.000 60.264 40.000 30.462 50.803 40.247 60.303 10.049 70.514 40.000 50.558 20.000 30.111 10.000 20.556 10.000 20.000 10.406 40.536 20.681 10.484 20.346 30.925 30.470 20.664 20.726 10.130 90.780 40.000 10.009 50.618 50.764 90.487 60.000 10.442 30.245 100.000 10.593 10.655 40.345 70.411 40.279 3
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
ACGP-ScanNet2000.381 20.486 20.362 20.275 10.597 40.133 20.333 20.500 10.394 20.000 10.008 90.483 20.512 41.000 10.649 50.497 50.792 40.032 30.556 20.179 40.170 40.307 50.291 40.480 30.114 71.000 10.242 20.000 10.037 10.512 40.365 61.000 10.000 10.396 40.000 30.608 40.184 50.643 70.009 20.007 40.271 20.209 10.304 50.000 50.000 10.731 10.678 30.248 50.779 11.000 10.647 90.080 10.288 40.423 80.000 50.396 20.435 40.903 10.499 40.400 20.676 20.247 10.329 40.500 10.062 80.462 20.673 20.144 20.574 10.252 30.000 20.365 50.000 60.336 10.733 50.556 10.412 20.312 20.581 30.524 30.313 40.000 20.000 10.349 50.037 40.301 20.036 30.194 30.143 60.600 10.652 30.677 20.314 10.772 20.000 40.000 10.444 80.104 10.031 40.486 20.077 10.472 21.000 10.635 40.500 20.454 10.500 30.782 20.449 50.018 50.538 40.069 50.406 40.002 10.146 70.014 30.795 20.000 30.139 70.001 20.686 10.815 20.541 10.753 50.556 20.007 40.284 10.330 50.778 20.926 20.792 50.785 10.444 10.380 30.000 30.514 20.821 20.346 20.197 20.065 60.494 50.000 50.395 100.000 30.000 30.000 20.391 50.000 20.000 10.546 20.543 10.548 30.438 30.240 60.895 50.388 60.569 50.694 20.197 70.824 20.000 10.060 30.612 71.000 10.832 30.000 10.461 20.752 10.000 10.486 20.850 10.466 20.400 50.112 7
Rongkun Yang, Ye Zhang, Longguang Wang, Zhiheng Fu, Lian Xu, Yulan Guo: Beyond Context Bias: Adaptive Instance Placement for Robust 3D Instance Segmentation.
Mask3D Scannet2000.278 60.383 60.263 70.168 60.506 60.068 40.083 110.000 40.000 60.000 10.023 70.149 100.302 50.778 80.647 60.569 20.500 50.031 40.014 70.027 80.173 30.311 30.195 60.351 80.258 40.000 60.082 70.000 10.003 30.037 80.391 51.000 10.000 10.014 80.000 30.572 60.573 20.661 60.000 50.003 70.005 100.082 100.349 30.028 30.000 10.605 50.515 90.509 10.711 31.000 10.665 70.015 80.107 70.402 90.201 10.083 60.304 60.759 20.491 50.378 50.572 60.119 60.277 60.013 110.089 40.283 70.411 70.267 10.006 90.156 40.000 20.116 60.000 60.105 90.556 70.514 40.396 30.275 50.323 50.215 70.380 10.000 20.000 10.356 40.005 70.208 50.325 10.000 50.050 90.400 30.561 60.258 70.179 60.722 50.147 20.000 10.586 40.063 40.015 50.139 70.016 40.028 60.708 50.418 80.016 70.048 90.500 30.489 60.349 60.001 80.475 70.086 40.365 60.000 20.500 10.000 40.323 90.000 30.222 40.000 30.497 50.626 50.044 90.795 30.556 20.008 30.121 100.265 60.667 30.789 50.568 60.579 40.444 10.176 60.004 20.474 30.752 60.233 70.014 60.002 100.570 30.007 20.377 110.000 30.000 30.000 20.337 70.000 20.000 10.384 50.465 50.287 70.085 60.048 80.816 110.467 30.810 10.377 70.415 10.744 50.000 10.004 60.724 10.778 60.590 40.000 10.032 70.441 50.000 10.377 60.391 50.427 50.321 60.192 4
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
AQ3D-ScanNet2000.385 10.522 10.374 10.234 30.698 10.127 30.500 10.500 10.574 10.000 10.023 60.383 50.549 21.000 10.687 30.527 40.889 20.013 50.533 50.239 10.148 50.458 10.358 10.492 20.147 60.167 50.209 30.000 10.009 20.548 30.469 10.167 70.000 10.451 20.000 30.641 20.073 60.759 20.009 30.005 50.299 10.203 30.296 60.037 20.000 10.399 60.683 20.139 70.676 51.000 10.754 10.030 60.306 30.670 10.150 30.025 80.507 10.020 70.565 30.570 10.671 30.224 40.520 20.333 20.077 60.451 30.733 10.141 30.292 60.156 40.000 20.424 40.337 10.276 30.748 40.444 60.435 10.304 30.598 20.554 20.363 20.000 20.000 10.468 30.075 30.341 10.000 50.438 20.125 70.400 30.585 40.658 40.231 50.771 30.000 40.000 10.655 10.052 50.303 20.556 10.066 20.667 11.000 10.801 10.083 50.267 51.000 10.777 30.578 10.028 20.681 20.000 70.479 20.000 20.140 80.000 40.667 40.000 30.444 20.132 10.501 40.856 10.475 20.799 20.556 20.029 20.221 40.422 11.000 10.928 11.000 10.482 70.444 10.454 20.000 30.462 40.792 50.331 30.158 30.101 30.773 10.000 50.455 40.000 20.000 30.000 20.398 40.000 20.000 10.532 30.510 30.617 20.395 40.367 20.923 40.448 50.657 30.590 30.410 20.825 10.000 10.083 20.704 21.000 10.864 10.000 10.664 10.618 30.000 10.333 80.667 30.406 60.455 20.551 1
Volt-SPFormerpermissive0.367 30.475 30.359 30.248 20.635 30.051 60.333 20.000 40.125 30.000 10.029 50.345 60.528 31.000 10.663 40.400 70.389 70.012 60.556 20.235 20.407 20.240 70.308 30.550 10.380 20.250 30.193 40.000 10.000 40.439 60.416 41.000 10.000 10.254 50.000 30.609 30.638 10.678 40.000 50.004 60.113 50.144 50.333 40.028 30.000 10.719 20.685 10.139 70.682 41.000 10.689 40.052 30.247 50.470 50.000 50.304 30.484 20.000 80.588 20.378 50.736 10.241 30.663 10.066 100.299 10.717 10.660 30.000 50.466 30.156 40.000 20.500 30.278 30.230 50.831 10.556 10.365 40.192 60.822 10.565 10.318 30.111 10.000 10.533 10.013 50.232 40.000 50.778 10.112 80.400 30.693 20.588 50.284 30.684 60.000 40.000 10.556 50.050 60.008 60.333 60.029 30.278 41.000 10.748 20.500 20.340 21.000 10.787 10.575 20.013 60.527 50.017 60.502 10.000 20.500 10.167 10.650 50.000 30.222 40.000 30.549 20.655 40.238 30.799 10.556 20.002 50.170 60.348 40.250 40.873 31.000 10.652 30.444 10.551 10.000 30.524 10.821 30.329 40.000 70.117 20.383 90.014 10.417 90.000 30.000 30.000 20.469 20.000 20.000 10.552 10.494 40.515 50.710 10.388 10.928 20.524 10.537 60.560 40.167 80.817 30.000 10.019 40.682 31.000 10.864 10.000 10.099 50.664 20.000 10.395 40.764 20.442 40.418 30.117 5
Kadir Yilmaz, Adrian Kruse, Tristan Höfer, Daan de Geus, Bastian Leibe: Volume Transformer: Revisiting Vanilla Transformers for 3D Scene Understanding.
TD3D Scannet200permissive0.211 80.332 80.177 80.103 80.337 80.036 80.222 100.000 40.000 60.000 10.031 40.342 70.093 100.852 50.452 100.559 30.000 80.004 70.000 80.039 60.000 60.309 40.047 100.380 60.028 80.000 60.080 80.000 10.000 40.147 70.192 90.000 80.000 10.083 70.000 30.395 70.039 100.662 50.000 50.000 80.074 60.135 60.296 60.000 50.000 10.231 100.646 40.139 70.633 81.000 10.705 30.048 50.088 80.439 60.184 20.039 70.266 70.551 40.260 90.026 110.463 80.046 90.252 70.249 50.083 50.372 60.411 60.000 50.414 40.323 10.000 20.052 70.000 60.157 70.278 80.278 80.237 80.015 80.321 60.253 60.060 100.000 20.000 10.272 70.008 60.169 60.032 40.000 50.404 10.356 60.283 80.073 90.028 110.617 70.038 30.000 10.494 60.037 70.215 30.083 80.000 60.003 80.486 80.694 30.000 80.040 100.083 100.219 110.209 80.007 70.483 60.000 70.125 90.000 20.150 60.014 20.544 70.000 30.000 80.000 30.260 100.143 110.200 50.610 80.028 70.032 10.145 70.059 80.046 80.740 70.806 40.543 60.000 60.108 80.008 10.222 110.669 70.456 10.074 50.224 10.586 20.006 30.451 50.000 30.002 20.889 10.282 80.000 20.000 10.252 80.413 70.111 80.074 70.240 70.893 60.266 80.144 90.293 80.281 30.604 80.000 10.000 70.379 110.963 50.250 100.000 10.160 40.420 70.000 10.343 70.207 70.079 110.315 70.052 8
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
ODIN - Ins200permissive0.265 70.349 70.268 60.163 70.360 70.054 50.278 60.000 40.125 30.000 10.031 30.506 10.266 70.630 90.609 70.481 60.903 10.000 81.000 10.032 70.000 60.022 100.138 70.314 100.310 30.000 60.178 50.000 10.000 40.552 20.421 30.889 60.000 10.451 20.097 10.357 80.054 80.485 110.052 10.040 20.210 30.160 40.370 10.000 50.000 10.191 110.529 70.250 40.617 91.000 10.492 110.016 70.197 60.324 100.000 50.250 50.265 80.167 50.317 70.200 80.549 70.107 70.231 80.119 90.141 30.253 80.267 80.000 50.565 20.111 80.000 20.000 80.278 30.285 20.665 60.389 70.306 70.077 70.037 110.186 110.156 80.000 20.000 10.478 20.000 80.091 80.204 20.000 50.345 20.200 80.550 70.674 30.160 70.526 80.438 10.000 10.476 70.035 80.003 70.444 30.000 60.333 30.361 90.606 50.083 50.332 30.417 80.327 70.297 70.035 10.615 30.281 30.083 100.000 20.250 40.000 40.610 60.000 30.333 30.000 30.238 110.481 70.218 40.440 101.000 10.000 60.229 30.257 70.000 90.746 60.361 110.188 80.000 60.221 50.000 30.320 70.655 80.193 80.000 70.067 50.389 80.000 50.594 10.037 10.000 30.000 20.371 60.000 20.000 10.344 70.366 90.506 60.074 70.250 50.848 90.451 40.389 70.546 50.205 50.698 70.000 10.000 70.494 90.769 80.493 50.000 10.000 80.463 40.000 10.333 80.333 60.640 10.251 80.115 6
Minkowski 34D Inst.permissive0.130 100.246 100.083 100.043 110.299 100.000 110.278 60.000 40.000 60.000 10.022 80.175 90.122 80.537 100.521 80.400 70.000 80.000 80.000 80.008 90.000 60.048 90.076 90.182 110.000 100.000 60.022 100.000 10.000 40.000 90.141 110.000 80.000 10.000 90.000 30.210 100.063 70.547 100.000 50.000 80.000 110.100 80.026 110.000 50.000 10.241 90.488 100.000 100.564 111.000 10.672 60.000 90.021 100.486 30.000 50.000 90.067 100.000 80.194 110.033 100.415 100.026 100.025 110.271 30.004 100.094 110.142 110.000 50.000 100.111 80.000 20.000 80.000 60.088 100.083 110.278 80.110 100.000 100.082 100.199 100.137 90.000 20.000 10.000 90.000 80.041 100.000 50.000 50.308 30.067 90.280 90.016 100.101 90.373 110.000 40.000 10.319 100.007 100.000 90.000 90.000 60.000 90.028 110.355 110.000 80.101 70.444 70.289 80.114 110.000 90.394 90.000 70.032 110.000 20.000 90.000 40.201 110.000 30.000 80.000 30.384 70.248 100.000 110.529 90.000 80.000 60.133 90.020 110.089 60.720 90.500 90.099 100.000 60.000 110.000 30.238 100.334 110.190 90.000 70.000 110.317 110.000 50.472 30.000 30.000 30.000 20.094 110.000 20.000 10.082 110.236 100.004 110.019 100.000 90.883 70.061 110.262 80.217 100.000 100.557 110.000 10.000 70.460 100.761 100.156 110.000 10.000 80.259 90.000 10.394 50.019 100.084 100.232 100.000 11
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
CSC-Pretrain Inst.permissive0.123 110.223 110.082 110.046 100.308 90.004 90.278 60.000 40.000 60.000 10.000 110.032 110.105 90.537 100.348 110.378 90.000 80.000 80.000 80.000 110.000 60.000 110.037 110.323 90.000 100.000 60.013 110.000 10.000 40.000 90.235 80.000 80.000 10.000 90.000 30.231 90.045 90.564 90.000 50.000 80.006 90.078 110.065 90.000 50.000 10.259 80.516 80.000 100.600 101.000 10.578 100.000 90.000 110.184 110.000 50.000 90.034 110.000 80.211 100.089 90.394 110.018 110.064 100.171 80.001 110.144 90.172 100.000 50.000 100.044 100.000 20.000 80.000 60.064 110.126 100.278 80.093 110.000 100.094 90.214 80.011 110.000 20.000 10.000 90.000 80.022 110.000 50.000 50.275 40.000 100.275 100.000 110.098 100.407 100.000 40.000 10.250 110.007 110.000 90.000 90.000 60.000 90.333 100.376 100.000 80.000 110.042 110.285 90.119 100.000 90.224 110.000 70.184 80.000 20.000 90.000 40.244 100.000 30.000 80.000 30.377 80.378 80.051 70.424 110.000 80.000 60.116 110.030 100.125 50.441 100.444 100.063 110.000 60.042 90.000 30.297 80.483 90.096 110.000 70.028 80.338 100.000 50.444 60.000 30.000 30.000 20.189 100.000 20.000 10.141 100.152 110.017 100.000 110.000 90.838 100.193 90.111 110.105 110.198 60.588 90.000 10.000 70.542 80.343 110.267 90.000 10.000 80.108 110.000 10.333 80.000 110.228 80.202 110.022 10
Ji Hou, Benjamin Graham, Matthias Nießner, Saining Xie: Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts. CVPR 2021
LGround Inst.permissive0.154 90.275 90.108 90.060 90.295 110.002 100.278 60.000 40.000 60.000 10.006 100.272 80.064 110.815 70.503 90.333 100.000 80.000 80.556 20.001 100.000 60.148 80.078 80.448 40.007 90.000 60.024 90.000 10.000 40.000 90.190 100.000 80.000 10.000 90.000 30.209 110.031 110.573 80.000 50.000 80.041 80.099 90.037 100.000 50.000 10.327 70.364 110.181 60.642 71.000 10.654 80.000 90.023 90.429 70.000 50.000 90.097 90.000 80.278 80.267 70.434 90.048 80.092 90.257 40.030 90.097 100.189 90.000 50.089 70.000 110.000 20.000 80.000 60.115 80.166 90.222 110.222 90.003 90.127 70.213 90.169 70.000 20.000 10.000 90.000 80.044 90.000 50.000 50.000 100.000 100.268 110.222 80.130 80.494 90.000 40.000 10.363 90.015 90.000 90.000 90.000 60.000 90.611 70.400 90.000 80.056 80.278 90.242 100.180 90.000 90.383 100.000 70.209 70.000 20.000 90.000 40.364 80.000 30.000 80.000 30.323 90.302 90.019 100.654 70.000 80.000 60.141 80.045 90.000 90.427 110.514 70.143 90.000 60.028 100.000 30.252 90.402 100.156 100.000 70.028 80.470 70.000 50.444 60.000 30.000 30.000 20.205 90.000 20.000 10.203 90.381 80.026 90.037 90.000 90.881 80.099 100.135 100.239 90.000 100.585 100.000 10.000 70.616 60.778 60.322 80.000 10.000 80.407 80.000 10.333 80.148 80.177 90.242 90.028 9
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild.