3D Semantic Instance Results - ScanNet Benchmark

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

Evaluation and metrics

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). Note that multiple predictions of the same ground truth instance are penalized as false positives.

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

Method	avg ap	bathtub	bed	bookshelf	cabinet	chair	counter	curtain	desk	door	otherfurniture	picture	refrigerator	shower curtain	sink	sofa	table	toilet	window

PointComp	0.629 1	0.787 25	0.679 10	0.574 5	0.502 3	0.824 1	0.378 1	0.480 39	0.483 3	0.480 16	0.601 1	0.744 1	0.682 8	0.809 8	0.460 21	0.819 1	0.643 2	0.935 13	0.449 3

PointRel	0.622 2	0.926 8	0.710 3	0.541 11	0.502 2	0.772 8	0.314 5	0.598 11	0.425 10	0.504 11	0.565 3	0.650 8	0.716 2	0.809 7	0.476 12	0.747 6	0.618 3	0.963 4	0.364 21
: Relation3D: Enhancing Relation Modeling for Point Cloud Instance Segmentation. CVPR 2025
Competitor-MAFT	0.618 3	0.866 16	0.724 1	0.628 1	0.484 5	0.803 3	0.300 9	0.509 32	0.496 1	0.539 1	0.547 7	0.703 2	0.668 9	0.708 34	0.463 18	0.708 18	0.595 5	0.959 6	0.418 9

SIM3D	0.617 4	0.952 4	0.629 19	0.539 12	0.426 17	0.768 12	0.302 8	0.681 2	0.425 11	0.473 18	0.511 17	0.701 3	0.717 1	0.821 6	0.467 15	0.774 2	0.559 16	0.914 20	0.448 4

Spherical Mask(CtoF)	0.616 5	0.946 5	0.654 14	0.555 7	0.434 14	0.769 11	0.271 14	0.604 8	0.447 6	0.505 9	0.549 4	0.698 4	0.716 2	0.775 17	0.480 9	0.747 7	0.575 12	0.925 15	0.436 6

EV3D	0.615 6	0.946 5	0.652 15	0.555 7	0.433 15	0.773 7	0.271 15	0.604 8	0.447 6	0.506 8	0.544 8	0.698 4	0.716 2	0.775 17	0.480 9	0.747 7	0.572 14	0.925 15	0.435 7

DCD	0.614 7	0.892 13	0.633 18	0.434 30	0.495 4	0.810 2	0.292 10	0.501 33	0.408 12	0.525 5	0.582 2	0.688 6	0.625 11	0.801 9	0.608 1	0.672 22	0.649 1	0.965 3	0.476 1

ExtMask3D	0.598 8	0.852 17	0.692 8	0.433 33	0.461 9	0.791 5	0.264 16	0.488 36	0.493 2	0.508 7	0.528 16	0.594 14	0.706 6	0.791 11	0.483 7	0.734 11	0.595 6	0.911 22	0.437 5

MAFT	0.596 9	0.889 14	0.721 2	0.448 25	0.460 10	0.768 13	0.251 18	0.558 21	0.408 13	0.504 10	0.539 10	0.616 12	0.618 13	0.858 3	0.482 8	0.684 21	0.551 19	0.931 14	0.450 2

UniPerception	0.588 10	0.963 3	0.667 12	0.493 16	0.472 8	0.750 17	0.229 21	0.528 27	0.468 5	0.498 14	0.542 9	0.643 9	0.530 23	0.661 41	0.463 17	0.695 20	0.599 4	0.972 1	0.420 8

MG-Former	0.587 11	0.852 17	0.639 17	0.454 24	0.393 23	0.758 16	0.338 3	0.572 16	0.480 4	0.527 3	0.491 24	0.671 7	0.527 24	0.867 1	0.485 6	0.601 33	0.590 9	0.938 12	0.390 13

InsSSM	0.586 12	1.000 1	0.593 23	0.440 28	0.480 6	0.771 9	0.345 2	0.437 42	0.444 9	0.495 15	0.548 6	0.579 18	0.621 12	0.720 30	0.409 25	0.712 13	0.593 7	0.960 5	0.395 11
Lei Yao, Yi Wang, Moyun Liu, Lap-Pui Chau: SGIFormer: Semantic-guided and Geometric-enhanced Interleaving Transformer for 3D Instance Segmentation. TCSVT, 2024
Queryformer	0.583 13	0.926 8	0.702 5	0.393 39	0.504 1	0.733 23	0.276 13	0.527 28	0.373 19	0.479 17	0.534 12	0.533 25	0.697 7	0.720 31	0.436 23	0.745 9	0.592 8	0.958 7	0.363 22

KmaxOneFormerNet	0.581 14	0.745 30	0.692 9	0.551 9	0.458 11	0.798 4	0.264 17	0.531 26	0.369 21	0.513 6	0.531 15	0.632 10	0.494 27	0.798 10	0.567 3	0.648 26	0.558 18	0.950 9	0.362 24

Competitor-SPFormer	0.580 15	0.721 37	0.705 4	0.593 4	0.444 13	0.786 6	0.286 11	0.564 19	0.376 18	0.498 13	0.534 13	0.546 23	0.390 47	0.785 13	0.577 2	0.708 17	0.579 11	0.954 8	0.388 14

VDG-Uni3DSeg	0.576 16	0.833 21	0.699 6	0.483 18	0.412 21	0.767 14	0.313 6	0.461 41	0.446 8	0.526 4	0.498 22	0.584 15	0.551 19	0.743 26	0.464 16	0.766 3	0.538 23	0.919 18	0.363 23

PBNet	0.573 17	0.926 8	0.575 29	0.619 2	0.472 7	0.736 21	0.239 20	0.487 37	0.383 17	0.459 21	0.506 20	0.533 24	0.585 15	0.767 19	0.404 26	0.717 12	0.559 17	0.969 2	0.381 17
Weiguang Zhao, Yuyao Yan, Chaolong Yang, Jianan Ye, Xi Yang, Kaizhu Huang: Divide and Conquer: 3D Instance Segmentation With Point-Wise Binarization. ICCV 2023
TST3D	0.569 18	0.778 27	0.675 11	0.598 3	0.451 12	0.727 24	0.280 12	0.476 40	0.395 14	0.472 19	0.457 30	0.583 16	0.580 17	0.777 14	0.462 20	0.735 10	0.547 21	0.919 19	0.333 30
Duc Tran Dang Trung, Byeongkeun Kang, Yeejin Lee: MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation. ACM Multimedia 2024
Mask3D	0.566 19	0.926 8	0.597 22	0.408 36	0.420 19	0.737 20	0.239 19	0.598 11	0.386 16	0.458 22	0.549 4	0.568 21	0.716 2	0.601 47	0.480 9	0.646 27	0.575 12	0.922 17	0.364 20
Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, Bastian Leibe: Mask3D for 3D Semantic Instance Segmentation. ICRA 2023
OneFormer3D	0.566 19	0.781 26	0.697 7	0.562 6	0.431 16	0.770 10	0.331 4	0.400 48	0.373 20	0.529 2	0.504 21	0.568 20	0.475 31	0.732 28	0.470 13	0.762 4	0.550 20	0.871 37	0.379 18
Maxim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: OneFormer3D: One Transformer for Unified Point Cloud Segmentation.
ISBNet	0.559 21	0.939 7	0.655 13	0.383 42	0.426 18	0.763 15	0.180 23	0.534 25	0.386 15	0.499 12	0.509 19	0.621 11	0.427 41	0.704 36	0.467 14	0.649 25	0.571 15	0.948 10	0.401 10
Tuan Duc Ngo, Binh-Son Hua, Khoi Nguyen: ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution. CVPR 2023
GraphCut	0.552 22	1.000 1	0.611 21	0.438 29	0.392 24	0.714 25	0.139 27	0.598 13	0.327 25	0.389 25	0.510 18	0.598 13	0.427 42	0.754 22	0.463 19	0.761 5	0.588 10	0.903 25	0.329 32

SPFormer	0.549 23	0.745 30	0.640 16	0.484 17	0.395 22	0.739 19	0.311 7	0.566 18	0.335 23	0.468 20	0.492 23	0.555 22	0.478 30	0.747 24	0.436 22	0.712 14	0.540 22	0.893 29	0.343 29
Sun Jiahao, Qing Chunmei, Tan Junpeng, Xu Xiangmin: Superpoint Transformer for 3D Scene Instance Segmentation. AAAI 2023 [Oral]
DKNet	0.532 24	0.815 22	0.624 20	0.517 13	0.377 26	0.749 18	0.107 29	0.509 31	0.304 27	0.437 23	0.475 25	0.581 17	0.539 21	0.775 16	0.339 32	0.640 29	0.506 26	0.901 26	0.385 16
Yizheng Wu, Min Shi, Shuaiyuan Du, Hao Lu, Zhiguo Cao, Weicai Zhong: 3D Instances as 1D Kernels. ECCV 2022
IPCA-Inst	0.520 25	0.889 14	0.551 33	0.548 10	0.418 20	0.665 35	0.064 38	0.585 14	0.260 35	0.277 40	0.471 27	0.500 26	0.644 10	0.785 12	0.369 28	0.591 37	0.511 24	0.878 34	0.362 25

SoftGroup++	0.513 26	0.704 39	0.578 28	0.398 38	0.363 32	0.704 26	0.061 39	0.647 5	0.297 32	0.378 28	0.537 11	0.343 30	0.614 14	0.828 5	0.295 37	0.710 16	0.505 28	0.875 36	0.394 12

SSTNet	0.506 27	0.738 34	0.549 34	0.497 15	0.316 38	0.693 29	0.178 24	0.377 52	0.198 41	0.330 31	0.463 29	0.576 19	0.515 25	0.857 4	0.494 4	0.637 30	0.457 32	0.943 11	0.290 41
Zhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia: Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks. ICCV2021
SoftGroup	0.504 28	0.667 46	0.579 26	0.372 44	0.381 25	0.694 28	0.072 35	0.677 3	0.303 28	0.387 26	0.531 14	0.319 34	0.582 16	0.754 21	0.318 33	0.643 28	0.492 29	0.907 24	0.388 15
Thang Vu, Kookhoi Kim, Tung M. Luu, Xuan Thanh Nguyen, Chang D. Yoo: SoftGroup for 3D Instance Segmentaiton on Point Clouds. CVPR 2022 [Oral]
DANCENET	0.504 28	0.926 8	0.579 25	0.472 20	0.367 29	0.626 45	0.165 25	0.432 43	0.221 37	0.408 24	0.449 32	0.411 28	0.564 18	0.746 25	0.421 24	0.707 19	0.438 35	0.846 45	0.288 42

TD3D	0.489 30	0.852 17	0.511 43	0.434 31	0.322 37	0.735 22	0.101 32	0.512 30	0.355 22	0.349 30	0.468 28	0.283 38	0.514 26	0.676 40	0.268 42	0.671 23	0.510 25	0.908 23	0.329 33
Maksim Kolodiazhnyi, Anna Vorontsova, Anton Konushin, Danila Rukhovich: Top-Down Beats Bottom-Up in 3D Instance Segmentation. WACV 2024
OccuSeg+instance	0.486 31	0.802 24	0.536 36	0.428 34	0.369 28	0.702 27	0.205 22	0.331 57	0.301 29	0.379 27	0.474 26	0.327 31	0.437 36	0.862 2	0.485 5	0.601 34	0.394 43	0.846 47	0.273 45
Lei Han, Tian Zheng, Lan Xu, Lu Fang: OccuSeg: Occupancy-aware 3D Instance Segmentation. CVPR2020
TopoSeg	0.479 32	0.704 39	0.564 30	0.467 22	0.366 30	0.633 43	0.068 36	0.554 22	0.262 34	0.328 32	0.447 33	0.323 32	0.534 22	0.722 29	0.288 39	0.614 31	0.482 30	0.912 21	0.358 27

DualGroup	0.469 33	0.815 22	0.552 32	0.398 37	0.374 27	0.683 31	0.130 28	0.539 24	0.310 26	0.327 33	0.407 36	0.276 39	0.447 35	0.535 51	0.342 31	0.659 24	0.455 33	0.900 28	0.301 37

SSEC	0.465 34	0.667 46	0.578 27	0.502 14	0.362 33	0.641 42	0.035 48	0.605 7	0.291 33	0.323 34	0.451 31	0.296 36	0.417 45	0.677 39	0.245 46	0.501 55	0.506 27	0.900 27	0.366 19

ODIN - Ins	0.463 35	0.738 34	0.589 24	0.344 48	0.358 34	0.560 54	0.139 26	0.393 51	0.331 24	0.373 29	0.392 39	0.496 27	0.493 28	0.709 33	0.377 27	0.599 35	0.359 49	0.752 57	0.332 31
Ayush Jain, Pushkal Katara, Nikolaos Gkanatsios, Adam W. Harley, Gabriel Sarch, Kriti Aggarwal, Vishrav Chaudhary, Katerina Fragkiadaki: ODIN: A Single Model for 2D and 3D Segmentation. CVPR 2024
HAIS	0.457 36	0.704 39	0.561 31	0.457 23	0.364 31	0.673 32	0.046 47	0.547 23	0.194 42	0.308 35	0.426 34	0.288 37	0.454 34	0.711 32	0.262 43	0.563 45	0.434 37	0.889 31	0.344 28
Shaoyu Chen, Jiemin Fang, Qian Zhang, Wenyu Liu, Xinggang Wang: Hierarchical Aggregation for 3D Instance Segmentation. ICCV 2021
DD-UNet+Group	0.436 37	0.630 54	0.508 46	0.480 19	0.310 40	0.624 47	0.065 37	0.638 6	0.174 43	0.256 44	0.384 41	0.194 51	0.428 39	0.759 20	0.289 38	0.574 42	0.400 41	0.849 44	0.291 40
H. Liu, R. Liu, K. Yang, J. Zhang, K. Peng, R. Stiefelhagen: HIDA: Towards Holistic Indoor Understanding for the Visually Impaired via Semantic Instance Segmentation with a Wearable Solid-State LiDAR Sensor. ICCVW 2021
INS-Conv-instance	0.435 38	0.716 38	0.495 48	0.355 46	0.331 35	0.689 30	0.102 31	0.394 50	0.208 40	0.280 38	0.395 38	0.250 42	0.544 20	0.741 27	0.309 35	0.536 51	0.391 44	0.842 50	0.258 49

Mask-Group	0.434 39	0.778 27	0.516 41	0.471 21	0.330 36	0.658 36	0.029 50	0.526 29	0.249 36	0.256 43	0.400 37	0.309 35	0.384 50	0.296 67	0.368 29	0.575 41	0.425 38	0.877 35	0.362 26
Min Zhong, Xinghao Chen, Xiaokang Chen, Gang Zeng, Yunhe Wang: MaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation. ICME 2022
Box2Mask	0.433 40	0.741 32	0.463 53	0.433 32	0.283 43	0.625 46	0.103 30	0.298 62	0.125 52	0.260 42	0.424 35	0.322 33	0.472 32	0.701 37	0.363 30	0.711 15	0.309 61	0.882 32	0.272 47
Julian Chibane, Francis Engelmann, Tuan Anh Tran, Gerard Pons-Moll: Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes. ECCV 2022
RPGN	0.428 41	0.630 54	0.508 45	0.367 45	0.249 50	0.658 37	0.016 58	0.673 4	0.131 50	0.234 47	0.383 42	0.270 40	0.434 37	0.748 23	0.274 41	0.609 32	0.406 40	0.842 49	0.267 48
Shichao Dong, Guosheng Lin, Tzu-Yi Hung: Learning Regional Purity for Instance Segmentation on 3D Point Clouds. ECCV 2022
DENet	0.413 42	0.741 32	0.520 38	0.237 58	0.284 42	0.523 57	0.097 33	0.691 1	0.138 47	0.209 57	0.229 59	0.238 45	0.390 48	0.707 35	0.310 34	0.448 62	0.470 31	0.892 30	0.310 35

PointGroup	0.407 43	0.639 53	0.496 47	0.415 35	0.243 52	0.645 41	0.021 55	0.570 17	0.114 53	0.211 55	0.359 44	0.217 49	0.428 40	0.660 42	0.256 44	0.562 46	0.341 53	0.860 40	0.291 39
Li Jiang, Hengshuang Zhao, Shaoshuai Shi, Shu Liu, Chi-Wing Fu, Jiaya Jia: PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation. CVPR 2020 [oral]
CSC-Pretrained	0.405 44	0.738 34	0.465 52	0.331 51	0.205 56	0.655 38	0.051 43	0.601 10	0.092 57	0.211 56	0.329 47	0.198 50	0.459 33	0.775 15	0.195 53	0.524 53	0.400 42	0.878 33	0.184 58

PE	0.396 45	0.667 46	0.467 51	0.446 27	0.243 51	0.624 48	0.022 54	0.577 15	0.106 54	0.219 50	0.340 45	0.239 44	0.487 29	0.475 58	0.225 48	0.541 50	0.350 51	0.818 52	0.273 46
Biao Zhang, Peter Wonka: Point Cloud Instance Segmentation using Probabilistic Embeddings. CVPR 2021
Dyco3D	0.395 46	0.642 52	0.518 40	0.447 26	0.259 49	0.666 34	0.050 44	0.251 67	0.166 44	0.231 48	0.362 43	0.232 46	0.331 53	0.535 50	0.229 47	0.587 38	0.438 36	0.850 42	0.317 34
Tong He; Chunhua Shen; Anton van den Hengel: DyCo3D: Robust Instance Segmentation of 3D Point Clouds through Dynamic Convolution. CVPR2021
OSIS	0.392 47	0.778 27	0.530 37	0.220 60	0.278 44	0.567 53	0.083 34	0.330 58	0.299 30	0.270 41	0.310 50	0.143 57	0.260 57	0.624 45	0.277 40	0.568 44	0.361 48	0.865 39	0.301 36

AOIA	0.387 48	0.704 39	0.515 42	0.385 41	0.225 55	0.669 33	0.005 65	0.482 38	0.126 51	0.181 60	0.269 56	0.221 48	0.426 43	0.478 57	0.218 49	0.592 36	0.371 46	0.851 41	0.242 51

SSEN	0.384 49	0.852 17	0.494 49	0.192 61	0.226 54	0.648 40	0.022 53	0.398 49	0.299 31	0.277 39	0.317 49	0.231 47	0.194 64	0.514 54	0.196 51	0.586 39	0.444 34	0.843 48	0.184 57
Dongsu Zhang, Junha Chun, Sang Kyun Cha, Young Min Kim: Spatial Semantic Embedding Network: Fast 3D Instance Segmentation with Deep Metric Learning. Arxiv
Mask3D_evaluation	0.382 50	0.593 56	0.520 39	0.390 40	0.314 39	0.600 49	0.018 57	0.287 65	0.151 46	0.281 37	0.387 40	0.169 55	0.429 38	0.654 43	0.172 57	0.578 40	0.384 45	0.670 64	0.278 44

PCJC	0.375 51	0.704 39	0.542 35	0.284 55	0.197 58	0.649 39	0.006 62	0.426 44	0.138 48	0.242 45	0.304 51	0.183 54	0.388 49	0.629 44	0.141 64	0.546 49	0.344 52	0.738 59	0.283 43

ClickSeg_Instance	0.366 52	0.654 50	0.375 57	0.184 62	0.302 41	0.592 51	0.050 45	0.300 61	0.093 56	0.283 36	0.277 53	0.249 43	0.426 44	0.615 46	0.299 36	0.504 54	0.367 47	0.832 51	0.191 56

SphereSeg	0.357 53	0.651 51	0.411 55	0.345 47	0.264 48	0.630 44	0.059 40	0.289 64	0.212 38	0.240 46	0.336 46	0.158 56	0.305 54	0.557 48	0.159 60	0.455 61	0.341 54	0.726 61	0.294 38

3D-MPA	0.355 54	0.457 66	0.484 50	0.299 53	0.277 45	0.591 52	0.047 46	0.332 55	0.212 39	0.217 51	0.278 52	0.193 52	0.413 46	0.410 61	0.195 52	0.574 43	0.352 50	0.849 43	0.213 54
Francis Engelmann, Martin Bokeloh, Alireza Fathi, Bastian Leibe, Matthias Nießner: 3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation. CVPR 2020
NeuralBF	0.353 55	0.593 56	0.511 44	0.375 43	0.264 47	0.597 50	0.008 60	0.332 56	0.160 45	0.229 49	0.274 55	0.000 78	0.206 61	0.678 38	0.155 61	0.485 57	0.422 39	0.816 53	0.254 50
Weiwei Sun, Daniel Rebain, Renjie Liao, Vladimir Tankovich, Soroosh Yazdani, Kwang Moo Yi, Andrea Tagliasacchi: NeuralBF: Neural Bilateral Filtering for Top-down Instance Segmentation on Point Clouds. WACV 2023
RWSeg	0.348 56	0.475 63	0.456 54	0.320 52	0.275 46	0.476 59	0.020 56	0.491 35	0.056 64	0.212 54	0.320 48	0.261 41	0.302 55	0.520 52	0.182 55	0.557 47	0.285 63	0.867 38	0.197 55

GICN	0.341 57	0.580 58	0.371 58	0.344 49	0.198 57	0.469 60	0.052 42	0.564 20	0.093 55	0.212 53	0.212 61	0.127 59	0.347 52	0.537 49	0.206 50	0.525 52	0.329 56	0.729 60	0.241 52

One_Thing_One_Click	0.326 58	0.472 64	0.361 59	0.232 59	0.183 59	0.555 55	0.000 71	0.498 34	0.038 66	0.195 58	0.226 60	0.362 29	0.168 65	0.469 59	0.251 45	0.553 48	0.335 55	0.846 46	0.117 66
Zhengzhe Liu, Xiaojuan Qi, Chi-Wing Fu: One Thing One Click: A Self-Training Approach for Weakly Supervised 3D Semantic Segmentation. CVPR 2021
Occipital-SCS	0.320 59	0.679 45	0.352 60	0.334 50	0.229 53	0.436 61	0.025 51	0.412 47	0.058 62	0.161 65	0.240 58	0.085 61	0.262 56	0.496 56	0.187 54	0.467 59	0.328 57	0.775 54	0.231 53

Sparse R-CNN	0.292 60	0.704 39	0.213 70	0.153 64	0.154 61	0.551 56	0.053 41	0.212 68	0.132 49	0.174 62	0.274 54	0.070 63	0.363 51	0.441 60	0.176 56	0.424 64	0.234 65	0.758 56	0.161 62

MTML	0.282 61	0.577 59	0.380 56	0.182 63	0.107 67	0.430 62	0.001 68	0.422 45	0.057 63	0.179 61	0.162 64	0.070 64	0.229 59	0.511 55	0.161 58	0.491 56	0.313 58	0.650 67	0.162 60
Jean Lahoud, Bernard Ghanem, Marc Pollefeys, Martin R. Oswald: 3D Instance Segmentation via Multi-task Metric Learning. ICCV 2019 [oral]
SALoss-ResNet	0.262 62	0.667 46	0.335 61	0.067 71	0.123 65	0.427 63	0.022 52	0.280 66	0.058 61	0.216 52	0.211 62	0.039 67	0.142 67	0.519 53	0.106 68	0.338 68	0.310 60	0.721 62	0.138 63
Zhidong Liang, Ming Yang, Hao Li, Chunxiang Wang: 3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud Segmentation. IEEE Robotics and Automation Letters (IROS2020)
MASC	0.254 63	0.463 65	0.249 69	0.113 65	0.167 60	0.412 65	0.000 70	0.374 53	0.073 58	0.173 63	0.243 57	0.130 58	0.228 60	0.368 63	0.160 59	0.356 66	0.208 66	0.711 63	0.136 64
Chen Liu, Yasutaka Furukawa: MASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation.
3D-BoNet	0.253 64	0.519 61	0.324 64	0.251 57	0.137 64	0.345 70	0.031 49	0.419 46	0.069 59	0.162 64	0.131 66	0.052 65	0.202 63	0.338 65	0.147 63	0.301 71	0.303 62	0.651 66	0.178 59
Bo Yang, Jianan Wang, Ronald Clark, Qingyong Hu, Sen Wang, Andrew Markham, Niki Trigoni: Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds. NeurIPS 2019 Spotlight
SPG_WSIS	0.251 65	0.380 68	0.274 67	0.289 54	0.144 62	0.413 64	0.000 71	0.311 59	0.065 60	0.113 67	0.130 67	0.029 70	0.204 62	0.388 62	0.108 67	0.459 60	0.311 59	0.769 55	0.127 65

SegGroup_ins	0.246 66	0.556 60	0.335 62	0.062 73	0.115 66	0.490 58	0.000 71	0.297 63	0.018 70	0.186 59	0.142 65	0.083 62	0.233 58	0.216 69	0.153 62	0.469 58	0.251 64	0.744 58	0.083 69
An Tao, Yueqi Duan, Yi Wei, Jiwen Lu, Jie Zhou: SegGroup: Seg-Level Supervision for 3D Instance and Semantic Segmentation. TIP 2022
PanopticFusion-inst	0.214 67	0.250 73	0.330 63	0.275 56	0.103 68	0.228 76	0.000 71	0.345 54	0.024 68	0.088 69	0.203 63	0.186 53	0.167 66	0.367 64	0.125 65	0.221 74	0.112 76	0.666 65	0.162 61
Gaku Narita, Takashi Seno, Tomoya Ishikawa, Yohsuke Kaji: PanopticFusion: Online Volumetric Semantic Mapping at the Level of Stuff and Things. IROS 2019 (to appear)
UNet-backbone	0.161 68	0.519 61	0.259 68	0.084 67	0.059 70	0.325 72	0.002 66	0.093 73	0.009 72	0.077 71	0.064 70	0.045 66	0.044 74	0.161 71	0.045 70	0.331 69	0.180 68	0.566 68	0.033 78

3D-SIS	0.161 68	0.407 67	0.155 75	0.068 70	0.043 74	0.346 69	0.001 67	0.134 70	0.005 73	0.088 68	0.106 69	0.037 68	0.135 69	0.321 66	0.028 74	0.339 67	0.116 75	0.466 71	0.093 68
Ji Hou, Angela Dai, Matthias Niessner: 3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans. CVPR 2019
R-PointNet	0.158 70	0.356 69	0.173 73	0.113 66	0.140 63	0.359 66	0.012 59	0.023 76	0.039 65	0.134 66	0.123 68	0.008 74	0.089 70	0.149 72	0.117 66	0.221 73	0.128 73	0.563 69	0.094 67

Region-18class	0.146 71	0.175 77	0.321 65	0.080 68	0.062 69	0.357 67	0.000 71	0.307 60	0.002 75	0.066 72	0.044 72	0.000 78	0.018 76	0.036 77	0.054 69	0.447 63	0.133 71	0.472 70	0.060 73

SemRegionNet-20cls	0.121 72	0.296 71	0.203 71	0.071 69	0.058 71	0.349 68	0.000 71	0.150 69	0.019 69	0.054 74	0.034 75	0.017 73	0.052 72	0.042 76	0.013 77	0.209 75	0.183 67	0.371 72	0.057 74

3D-BEVIS	0.117 73	0.250 73	0.308 66	0.020 77	0.009 79	0.269 75	0.006 63	0.008 77	0.029 67	0.037 77	0.014 78	0.003 76	0.036 75	0.147 73	0.042 72	0.381 65	0.118 74	0.362 73	0.069 72
Cathrin Elich, Francis Engelmann, Jonas Schult, Theodora Kontogianni, Bastian Leibe: 3D-BEVIS: Birds-Eye-View Instance Segmentation.
Hier3D	0.117 73	0.222 75	0.161 74	0.054 75	0.027 76	0.289 73	0.000 71	0.124 71	0.001 77	0.079 70	0.061 71	0.027 71	0.141 68	0.240 68	0.005 78	0.310 70	0.129 72	0.153 78	0.081 70
Tan: HCFS3D: Hierarchical Coupled Feature Selection Network for 3D Semantic and Instance Segmentation.
tmp	0.113 75	0.333 70	0.151 76	0.056 74	0.053 72	0.344 71	0.000 71	0.105 72	0.016 71	0.049 75	0.035 74	0.020 72	0.053 71	0.048 75	0.013 76	0.183 77	0.173 69	0.344 75	0.054 75

Sem_Recon_ins	0.098 76	0.295 72	0.187 72	0.015 78	0.036 75	0.213 77	0.005 64	0.038 75	0.003 74	0.056 73	0.037 73	0.036 69	0.015 77	0.051 74	0.044 71	0.209 76	0.098 77	0.354 74	0.071 71

ASIS	0.085 77	0.037 78	0.080 78	0.066 72	0.047 73	0.282 74	0.000 71	0.052 74	0.002 76	0.047 76	0.026 76	0.001 77	0.046 73	0.194 70	0.031 73	0.264 72	0.140 70	0.167 77	0.047 77

Sgpn_scannet	0.049 78	0.023 79	0.134 77	0.031 76	0.013 78	0.144 78	0.006 61	0.008 78	0.000 78	0.028 78	0.017 77	0.003 75	0.009 79	0.000 78	0.021 75	0.122 78	0.095 78	0.175 76	0.054 76

MaskRCNN 2d->3d Proj	0.022 79	0.185 76	0.000 79	0.000 79	0.015 77	0.000 79	0.000 69	0.006 79	0.000 78	0.010 79	0.006 79	0.107 60	0.012 78	0.000 78	0.002 79	0.027 79	0.004 79	0.022 79	0.001 79

3D Semantic Instance Benchmark