2D Semantic Label Results - ScanNet Benchmark

The 2D semantic labeling task involves predicting a per-pixel semantic labeling of an image.

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.

This table lists the benchmark results for the 2D semantic label scenario.

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

Virtual MVFusion (R)	0.745 1	0.861 1	0.839 1	0.881 1	0.672 1	0.512 1	0.422 10	0.898 1	0.723 1	0.714 1	0.954 2	0.454 1	0.509 1	0.773 1	0.895 1	0.756 1	0.820 1	0.653 1	0.935 1	0.891 1	0.728 1
Abhijit Kundu, Xiaoqi Yin, Alireza Fathi, David Ross, Brian Brewington, Thomas Funkhouser, Caroline Pantofaru: Virtual Multi-view Fusion for 3D Semantic Segmentation. ECCV 2020
CU-Hybrid-2D Net	0.636 3	0.825 2	0.820 2	0.179 16	0.648 3	0.463 3	0.549 1	0.742 3	0.676 2	0.628 2	0.961 1	0.420 2	0.379 5	0.684 3	0.381 10	0.732 2	0.723 3	0.599 2	0.827 8	0.851 2	0.634 3

BPNet_2D	0.670 2	0.822 3	0.795 3	0.836 2	0.659 2	0.481 2	0.451 6	0.769 2	0.656 3	0.567 3	0.931 3	0.395 3	0.390 4	0.700 2	0.534 3	0.689 5	0.770 2	0.574 3	0.865 3	0.831 3	0.675 2
Wenbo Hu, Hengshuang Zhao, Li Jiang, Jiaya Jia and Tien-Tsin Wong: Bidirectional Projection Network for Cross Dimension Scene Understanding. CVPR 2021 (Oral)
DMMF	0.597 5	0.543 11	0.755 6	0.749 4	0.585 6	0.338 6	0.494 4	0.704 6	0.598 4	0.494 10	0.911 6	0.347 8	0.327 9	0.593 7	0.527 4	0.675 7	0.646 8	0.513 4	0.842 6	0.774 8	0.527 11

DMMF_3d	0.605 4	0.651 6	0.744 7	0.782 3	0.637 4	0.387 4	0.536 2	0.732 4	0.590 5	0.540 4	0.856 13	0.359 6	0.306 10	0.596 6	0.539 2	0.627 12	0.706 4	0.497 6	0.785 12	0.757 11	0.476 13

MCA-Net	0.595 6	0.533 12	0.756 5	0.746 5	0.590 5	0.334 8	0.506 3	0.670 7	0.587 6	0.500 8	0.905 8	0.366 5	0.352 6	0.601 5	0.506 6	0.669 10	0.648 6	0.501 5	0.839 7	0.769 9	0.516 12

SSMA	0.577 9	0.695 4	0.716 10	0.439 12	0.563 8	0.314 10	0.444 8	0.719 5	0.551 7	0.503 7	0.887 10	0.346 9	0.348 7	0.603 4	0.353 12	0.709 3	0.600 11	0.457 10	0.901 2	0.786 6	0.599 7
Abhinav Valada, Rohit Mohan, Wolfram Burgard: Self-Supervised Model Adaptation for Multimodal Semantic Segmentation. International Journal of Computer Vision, 2019
DCRedNet	0.583 8	0.682 5	0.723 8	0.542 10	0.510 11	0.310 11	0.451 6	0.668 8	0.549 8	0.520 6	0.920 5	0.375 4	0.446 2	0.528 11	0.417 9	0.670 9	0.577 13	0.478 8	0.862 4	0.806 5	0.628 5

RFBNet	0.592 7	0.616 7	0.758 4	0.659 6	0.581 7	0.330 9	0.469 5	0.655 10	0.543 9	0.524 5	0.924 4	0.355 7	0.336 8	0.572 8	0.479 8	0.671 8	0.648 6	0.480 7	0.814 10	0.814 4	0.614 6

3DMV (2d proj)	0.498 13	0.481 15	0.612 14	0.579 9	0.456 13	0.343 5	0.384 12	0.623 13	0.525 10	0.381 14	0.845 14	0.254 13	0.264 14	0.557 9	0.182 15	0.581 15	0.598 12	0.429 13	0.760 14	0.661 16	0.446 15
Angela Dai, Matthias Niessner: 3DMV: Joint 3D-Multi-View Prediction for 3D Semantic Scene Segmentation. ECCV'18
SN_RN152pyrx8_RVC	0.546 10	0.572 9	0.663 13	0.638 8	0.518 9	0.298 12	0.366 15	0.633 12	0.510 11	0.446 12	0.864 11	0.296 11	0.267 12	0.542 10	0.346 13	0.704 4	0.575 14	0.431 12	0.853 5	0.766 10	0.630 4

FuseNet	0.535 11	0.570 10	0.681 12	0.182 15	0.512 10	0.290 13	0.431 9	0.659 9	0.504 12	0.495 9	0.903 9	0.308 10	0.428 3	0.523 12	0.365 11	0.676 6	0.621 10	0.470 9	0.762 13	0.779 7	0.541 9
Caner Hazirbas, Lingni Ma, Csaba Domokos, Daniel Cremers: FuseNet: Incorporating Depth into Semantic Segmentation via Fusion-based CNN Architecture. ACCV 2016
AdapNet++	0.503 12	0.613 8	0.722 9	0.418 13	0.358 17	0.337 7	0.370 14	0.479 15	0.443 13	0.368 15	0.907 7	0.207 14	0.213 16	0.464 15	0.525 5	0.618 13	0.657 5	0.450 11	0.788 11	0.721 14	0.408 16
Abhinav Valada, Rohit Mohan, Wolfram Burgard: Self-Supervised Model Adaptation for Multimodal Semantic Segmentation. International Journal of Computer Vision, 2019
ScanNet (2d proj)	0.330 17	0.293 16	0.521 16	0.657 7	0.361 16	0.161 16	0.250 16	0.004 17	0.440 14	0.183 17	0.836 15	0.125 16	0.060 17	0.319 17	0.132 16	0.417 16	0.412 16	0.344 16	0.541 17	0.427 17	0.109 17
Angela Dai, Angel X. Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, Matthias Nießner: ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes. CVPR'17
ILC-PSPNet	0.475 15	0.490 14	0.581 15	0.289 14	0.507 12	0.067 17	0.379 13	0.610 14	0.417 15	0.435 13	0.822 16	0.278 12	0.267 12	0.503 13	0.228 14	0.616 14	0.533 15	0.375 14	0.820 9	0.729 12	0.560 8

Enet (reimpl)	0.376 16	0.264 17	0.452 17	0.452 11	0.365 15	0.181 15	0.143 17	0.456 16	0.409 16	0.346 16	0.769 17	0.164 15	0.218 15	0.359 16	0.123 17	0.403 17	0.381 17	0.313 17	0.571 16	0.685 15	0.472 14
Re-implementation of Adam Paszke, Abhishek Chaurasia, Sangpil Kim, Eugenio Culurciello: ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation.
MSeg1080_RVC	0.485 14	0.505 13	0.709 11	0.092 17	0.427 14	0.241 14	0.411 11	0.654 11	0.385 17	0.457 11	0.861 12	0.053 17	0.279 11	0.503 13	0.481 7	0.645 11	0.626 9	0.365 15	0.748 15	0.725 13	0.529 10
John Lambert, Zhuang Liu, Ozan Sener, James Hays, Vladlen Koltun: MSeg: A Composite Dataset for Multi-domain Semantic Segmentation. CVPR 2020

2D Semantic label benchmark