The 3D semantic labeling task involves predicting a semantic labeling of a 3D scan mesh.

Evaluation and metrics

Our evaluation ranks all methods according to the PASCAL VOC intersection-over-union metric (IoU). IoU = TP/(TP+FP+FN), where TP, FP, and FN are the numbers of true positive, false positive, and false negative pixels, respectively. Predicted labels are evaluated per-vertex over the respective 3D scan mesh; for 3D approaches that operate on other representations like grids or points, the predicted labels should be mapped onto the mesh vertices (e.g., one such example for grid to mesh vertices is provided in the evaluation helpers).



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




Method Infoavg iouhead ioucommon ioutail iouwallchairfloortabledoorcouchcabinetshelfdeskoffice 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 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 by
Minkowski 34Dpermissive0.253 20.463 20.154 30.102 20.771 20.650 30.932 10.483 20.571 20.710 20.331 20.250 20.492 10.044 30.703 20.419 30.606 30.227 30.621 20.865 30.531 10.771 30.813 10.291 10.484 30.242 20.612 30.282 30.440 30.351 10.299 20.622 20.593 20.027 30.293 20.310 30.000 10.757 10.858 10.737 30.150 10.164 10.368 30.084 10.381 30.142 30.357 20.720 10.214 20.092 20.724 20.596 30.056 30.655 10.525 20.581 30.352 30.594 20.056 30.000 30.014 30.224 20.772 10.205 30.720 20.000 10.159 10.531 20.163 30.294 20.136 30.000 10.169 20.589 20.000 20.000 20.000 10.002 10.663 10.466 30.265 30.582 10.337 10.016 20.559 10.084 30.000 10.000 30.000 10.036 30.000 10.125 20.670 20.000 10.102 10.071 30.164 20.406 10.386 20.046 30.068 30.159 20.117 10.284 20.111 30.094 20.000 20.000 30.197 30.000 10.044 10.013 20.002 20.228 30.307 30.588 10.025 30.545 10.134 30.000 10.655 10.302 20.282 30.000 10.060 10.000 10.035 30.000 30.000 10.097 30.000 10.000 10.005 10.000 10.000 10.096 30.000 10.334 20.000 10.000 10.274 20.000 10.513 30.000 10.000 10.280 10.194 20.897 10.000 20.000 10.000 10.000 20.000 10.108 30.279 30.189 20.141 30.059 30.272 10.307 30.445 10.003 10.000 10.353 20.000 10.026 10.000 10.581 30.001 10.000 10.000 10.093 30.002 10.000 10.000 10.000 1
C. Choy, J. Gwak, S. Savarese: 4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks. CVPR 2019
LGroundpermissive0.272 10.485 10.184 10.106 10.778 10.676 10.932 10.479 30.572 10.718 10.399 10.265 10.453 20.085 20.745 10.446 10.726 10.232 20.622 10.901 10.512 20.826 10.786 20.178 30.549 20.277 10.659 20.381 10.518 10.295 30.323 10.777 10.599 10.028 20.321 10.363 20.000 10.708 20.858 10.746 20.063 20.022 20.457 10.077 20.476 10.243 10.402 10.397 30.233 10.077 30.720 30.610 20.103 10.629 20.437 30.626 10.446 10.702 10.190 10.005 10.058 20.322 10.702 20.244 10.768 10.000 10.134 30.552 10.279 20.395 10.147 20.000 10.207 10.612 10.000 20.000 20.000 10.000 20.658 20.566 10.323 20.525 30.229 20.179 10.467 30.154 20.000 10.002 10.000 10.051 10.000 10.127 10.703 10.000 10.000 20.216 10.112 30.358 20.547 10.187 10.092 20.156 30.055 30.296 10.252 10.143 10.000 20.014 10.398 20.000 10.028 20.173 10.000 30.265 20.348 10.415 30.179 10.019 20.218 10.000 10.597 20.274 30.565 10.000 10.012 30.000 10.039 20.022 20.000 10.117 10.000 10.000 10.000 20.000 10.000 10.324 20.000 10.384 10.000 10.000 10.251 30.000 10.566 10.000 10.000 10.066 20.404 10.886 20.199 10.000 10.000 10.059 10.000 10.136 10.540 10.127 30.295 10.085 20.143 30.514 10.413 30.000 20.000 10.498 10.000 10.000 20.000 10.623 10.000 20.000 10.000 10.132 20.000 20.000 10.000 10.000 1
David Rozenberszki, Or Litany, Angela Dai: Language-Grounded Indoor 3D Semantic Segmentation in the Wild. arXiv
CSC-Pretrainpermissive0.249 30.455 30.171 20.079 30.766 30.659 20.930 30.494 10.542 30.700 30.314 30.215 30.430 30.121 10.697 30.441 20.683 20.235 10.609 30.895 20.476 30.816 20.770 30.186 20.634 10.216 30.734 10.340 20.471 20.307 20.293 30.591 30.542 30.076 10.205 30.464 10.000 10.484 30.832 30.766 10.052 30.000 30.413 20.059 30.418 20.222 20.318 30.609 20.206 30.112 10.743 10.625 10.076 20.579 30.548 10.590 20.371 20.552 30.081 20.003 20.142 10.201 30.638 30.233 20.686 30.000 10.142 20.444 30.375 10.247 30.198 10.000 10.128 30.454 30.019 10.097 10.000 10.000 20.553 30.557 20.373 10.545 20.164 30.014 30.547 20.174 10.000 10.002 10.000 10.037 20.000 10.063 30.664 30.000 10.000 20.130 20.170 10.152 30.335 30.079 20.110 10.175 10.098 20.175 30.166 20.045 30.207 10.014 10.465 10.000 10.001 30.001 30.046 10.299 10.327 20.537 20.033 20.012 30.186 20.000 10.205 30.377 10.463 20.000 10.058 20.000 10.055 10.041 10.000 10.105 20.000 10.000 10.000 20.000 10.000 10.398 10.000 10.308 30.000 10.000 10.319 10.000 10.543 20.000 10.000 10.062 30.004 30.862 30.000 20.000 10.000 10.000 20.000 10.123 20.316 20.225 10.250 20.094 10.180 20.332 20.441 20.000 20.000 10.310 30.000 10.000 20.000 10.592 20.000 20.000 10.000 10.203 10.000 20.000 10.000 10.000 1
Ji Hou, Benjamin Graham, Matthias Nie├čner, Saining Xie: Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts. CVPR 2021