Benchmark Results - ScanRefer Benchmark

This table lists the benchmark results for the ScanRefer Localization Benchmark scenario.

		Unique	Unique	Multiple	Multiple	Overall	Overall
Method	Info	acc@0.25IoU	acc@0.5IoU	acc@0.25IoU	acc@0.5IoU	acc@0.25IoU	acc@0.5IoU
Method	Info
UniVLG		0.8895 1	0.8236 1	0.5921 1	0.5030 1	0.6588 1	0.5749 1
Ayush Jain, Alexander Swerdlow, Yuzhou Wang, Alexander Sax, Franziska Meier, Katerina Fragkiadaki: Unifying 2D and 3D Vision-Language Understanding.
Chat-Scene		0.8887 2	0.8005 2	0.5421 2	0.4861 2	0.6198 2	0.5566 2
Haifeng Huang, Yilun Chen, Zehan Wang, et al.: Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifiers. NeurIPS 2024
ConcreteNet		0.8607 3	0.7923 3	0.4746 9	0.4091 3	0.5612 8	0.4950 3
Ozan Unal, Christos Sakaridis, Suman Saha, Luc Van Gool: Four Ways to Improve Verbo-visual Fusion for Dense 3D Visual Grounding. ECCV 2024
M3DRef-SCLIP		0.7997 14	0.7123 4	0.4708 10	0.3805 10	0.5445 10	0.4549 8

M3DRef-CLIP		0.7980 15	0.7085 5	0.4692 11	0.3807 9	0.5433 11	0.4545 9
Yiming Zhang, ZeMing Gong, Angel X. Chang: Multi3DRefer: Grounding Text Description to Multiple 3D Objects. ICCV 2023
pointclip		0.8211 6	0.7082 6	0.4803 8	0.3884 7	0.5567 9	0.4601 7

cus3d		0.8384 5	0.7073 7	0.4908 7	0.4000 4	0.5688 6	0.4689 4

3DInsVG		0.8170 8	0.6925 8	0.4582 16	0.3617 11	0.5386 14	0.4359 11

D-LISA		0.8195 7	0.6900 9	0.4975 5	0.3967 6	0.5697 5	0.4625 5
Haomeng Zhang, Chiao-An Yang, Raymond A. Yeh: Multi-Object 3D Grounding with Dynamic Modules and Language-Informed Spatial Attention. NeurIPS 2024
CORE-3DVG		0.8557 4	0.6867 10	0.5275 3	0.3850 8	0.6011 3	0.4527 10

Clip-pre		0.7766 35	0.6843 11	0.3617 55	0.2904 41	0.4547 46	0.3787 29

D3Net		0.7923 21	0.6843 11	0.3905 39	0.3074 33	0.4806 38	0.3919 22
Dave Zhenyu Chen, Qirui Wu, Matthias Niessner, Angel X. Chang: D3Net: A Unified Speaker-Listener Architecture for 3D Dense Captioning and Visual Grounding. 17th European Conference on Computer Vision (ECCV), 2022
Clip		0.7733 38	0.6810 13	0.3619 53	0.2919 40	0.4542 47	0.3791 28

ContraRefer		0.7832 27	0.6801 14	0.3850 40	0.2947 35	0.4743 39	0.3811 25

M3DRef-test		0.7865 23	0.6793 15	0.4963 6	0.3977 5	0.5614 7	0.4608 6

RG-SAN		0.7964 16	0.6785 16	0.4591 15	0.3600 12	0.5348 17	0.4314 12

InstanceRefer		0.7782 34	0.6669 17	0.3457 59	0.2688 46	0.4427 52	0.3580 43
Zhihao Yuan, Xu Yan, Yinghong Liao, Ruimao Zhang, Zhen Li*, Shuguang Cui: InstanceRefer: Cooperative Holistic Understanding for Visual Grounding on Point Clouds through Instance Multi-level Contextual Referring. ICCV 2021
Se2d		0.7799 32	0.6628 18	0.3636 50	0.2823 43	0.4569 43	0.3677 37

3DVLP-rf		0.7964 16	0.6579 19	0.4646 13	0.3436 15	0.5390 12	0.4140 14

D3Net - Pretrained		0.7659 43	0.6579 19	0.3619 53	0.2726 44	0.4525 49	0.3590 41
Dave Zhenyu Chen, Qirui Wu, Matthias Niessner, Angel X. Chang: D3Net: A Unified Speaker-Listener Architecture for 3D Dense Captioning and Visual Grounding. 17th European Conference on Computer Vision (ECCV), 2022
3DVLP-rf-enhance		0.7939 19	0.6546 21	0.4651 12	0.3491 13	0.5388 13	0.4176 13

3dvlp-with-judge		0.7807 31	0.6472 22	0.4498 27	0.3407 16	0.5240 27	0.4094 16

PointGroup_MCAN		0.7510 46	0.6397 23	0.3271 62	0.2535 48	0.4222 56	0.3401 46

HAM		0.7799 32	0.6373 24	0.4148 35	0.3324 20	0.4967 35	0.4007 19
Jiaming Chen, Weixin Luo, Ran Song, Xiaolin Wei, Lin Ma, Wei Zhang: Learning Point-Language Hierarchical Alignment for 3D Visual Grounding.
3DVLP-baseline		0.7766 35	0.6373 24	0.4572 18	0.3469 14	0.5288 24	0.4120 15

Jung		0.8096 11	0.6331 26	0.5113 4	0.3398 18	0.5782 4	0.4055 17

ScanRefer-3dvlp-test		0.7824 29	0.6298 27	0.4532 25	0.3405 17	0.5270 26	0.4054 18

CSA-M3LM		0.8137 9	0.6241 28	0.4544 23	0.3317 21	0.5349 16	0.3972 20

secg		0.7288 53	0.6175 29	0.3696 48	0.2933 37	0.4501 50	0.3660 38

3DJCG(Grounding)		0.7675 42	0.6059 30	0.4389 30	0.3117 31	0.5126 30	0.3776 30
Daigang Cai, Lichen Zhao, Jing Zhang†, Lu Sheng, Dong Xu: 3DJCG: A Unified Framework for Joint Dense Captioning and Visual Grounding on 3D Point Clouds. CVPR2022 Oral
3dvlp-judge-h		0.7552 45	0.6051 31	0.4458 29	0.3340 19	0.5152 29	0.3948 21

TransformerVG		0.7502 47	0.5977 32	0.3712 47	0.2628 47	0.4562 45	0.3379 47

GALA-Grounder-D3		0.7939 19	0.5952 33	0.4625 14	0.3229 23	0.5368 15	0.3839 23

TGNN		0.6834 58	0.5894 34	0.3312 60	0.2526 49	0.4102 60	0.3281 49
Pin-Hao Huang, Han-Hung Lee, Hwann-Tzong Chen, Tyng-Luh Liu: Text-Guided Graph Neural Network for Referring 3D Instance Segmentation. AAAI 2021
HGT		0.7692 41	0.5886 35	0.4141 36	0.2924 39	0.4937 36	0.3588 42

FE-3DGQA		0.7857 24	0.5862 36	0.4317 32	0.2935 36	0.5111 32	0.3592 40

LAG-3D-3		0.7815 30	0.5837 37	0.4556 21	0.3219 24	0.5287 25	0.3806 26

Graph-VG-2		0.8021 13	0.5829 38	0.4546 22	0.3217 25	0.5325 19	0.3802 27

Graph-VG-3		0.8038 12	0.5812 39	0.4515 26	0.3169 27	0.5305 21	0.3762 31

LAG-3D-2		0.7964 16	0.5812 39	0.4572 18	0.3245 22	0.5333 18	0.3821 24

3DVG-Trans +		0.7733 38	0.5787 41	0.4370 31	0.3102 32	0.5124 31	0.3704 35
Lichen Zhao∗, Daigang Cai∗, Lu Sheng†, Dong Xu: 3DVG-Transformer: Relation Modeling for Visual Grounding on Point Clouds. ICCV2021
GALA-Grounder-D1		0.8104 10	0.5754 42	0.4479 28	0.3176 26	0.5292 23	0.3754 32

SAVG		0.7758 37	0.5664 43	0.4236 33	0.2826 42	0.5026 33	0.3462 45

SAF		0.6348 61	0.5647 44	0.3726 46	0.3009 34	0.4314 54	0.3601 39

Graph-VG-4		0.7848 25	0.5631 45	0.4560 20	0.3164 29	0.5298 22	0.3717 33

LAG-3D		0.7881 22	0.5606 46	0.4579 17	0.3169 27	0.5320 20	0.3715 34

bo3d-1		0.7469 49	0.5606 46	0.4539 24	0.3124 30	0.5196 28	0.3680 36

scanrefer-rj-14bz		0.7832 27	0.5524 48	0.3746 45	0.2275 53	0.4662 40	0.3004 53

3DVG-Transformer		0.7576 44	0.5515 49	0.4224 34	0.2933 37	0.4976 34	0.3512 44
Lichen Zhao∗, Daigang Cai∗, Lu Sheng†, Dong Xu: 3DVG-Transformer: Relation Modeling for Visual Grounding on Point Clouds. ICCV2021
BEAUTY-DETR		0.7848 25	0.5499 50	0.3934 38	0.2480 50	0.4811 37	0.3157 50
Ayush Jain, Nikolaos Gkanatsios, Ishita Mediratta, Katerina Fragkiadaki: Looking Outside the Box to Ground Language in 3D Scenes.
scanrefer-test-14bz		0.7700 40	0.5474 51	0.3665 49	0.2247 54	0.4569 43	0.2970 54

grounding		0.7298 52	0.5458 52	0.3822 42	0.2421 52	0.4538 48	0.3046 51

TFVG3D ++		0.7453 50	0.5458 53	0.3793 44	0.2690 45	0.4614 41	0.3311 48
Ali Solgi, Mehdi Ezoji: A Transformer-based Framework for Visual Grounding on 3D Point Clouds. AISP 2024
SR-GAB		0.7016 55	0.5202 54	0.3233 64	0.1959 59	0.4081 61	0.2686 57

henet		0.7110 54	0.5180 55	0.3936 37	0.2472 51	0.4590 42	0.3030 52

SRGA		0.7494 48	0.5128 56	0.3631 51	0.2218 55	0.4497 51	0.2871 55

scanrefer-rj-org		0.7345 51	0.4716 57	0.3536 57	0.2125 56	0.4390 53	0.2706 56

TransformerRefer		0.6010 63	0.4658 58	0.2540 70	0.1730 65	0.3318 70	0.2386 61

SPANet		0.5614 65	0.4641 59	0.2800 68	0.2071 58	0.3431 69	0.2647 58

ScanRefer-test		0.6999 56	0.4361 60	0.3274 61	0.1725 66	0.4109 59	0.2316 63

ScanRefer		0.6859 57	0.4353 61	0.3488 58	0.2097 57	0.4244 55	0.2603 59
Dave Zhenyu Chen, Angel X. Chang, Matthias Nießner: ScanRefer: 3D Object Localization in RGB-D Scans using Natural Language. 16th European Conference on Computer Vision (ECCV), 2020
scanrefer2		0.6340 62	0.4353 61	0.3193 65	0.1947 60	0.3898 63	0.2486 60

ScanRefer Baseline		0.6422 60	0.4196 63	0.3090 66	0.1832 61	0.3837 65	0.2362 62

ScanRefer_vanilla		0.6488 59	0.4056 64	0.3052 67	0.1782 63	0.3823 66	0.2292 64

pairwisemethod		0.5779 64	0.3603 65	0.2792 69	0.1746 64	0.3462 68	0.2163 65

bo3d		0.5400 66	0.1550 66	0.3817 43	0.1785 62	0.4172 58	0.1732 66

Co3d3		0.5326 67	0.1369 67	0.3848 41	0.1651 67	0.4179 57	0.1588 67

test_submitt		0.4732 70	0.1286 68	0.3626 52	0.1399 69	0.3874 64	0.1373 69

bo3d0		0.4823 69	0.1278 69	0.3271 62	0.1394 70	0.3619 67	0.1368 70

Co3d2		0.5070 68	0.1195 70	0.3569 56	0.1511 68	0.3906 62	0.1440 68

3DVLP		0.0038 71	0.0019 71	0.0049 71	0.0023 71	0.0047 71	0.0022 71

Co3d		0.0000 72	0.0000 72	0.0000 72	0.0000 72	0.0000 72	0.0000 72

		Captioning F1-Score				Dense Captioning	Object Detection
Method	Info	CIDEr@0.5IoU	BLEU-4@0.5IoU	Rouge-L@0.5IoU	METEOR@0.5IoU	DCmAP	mAP@0.5
Method	Info
Chat-Scene-thres0.5		0.3128 2	0.1679 4	0.2862 3	0.1376 2	0.1478 8	0.4981 4
Haifeng Huang, Yilun Chen, Zehan Wang, et al.: Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifiers. NeurIPS 2024
CM3D-Trans+		0.2348 6	0.1383 6	0.2250 7	0.1030 6	0.1398 9	0.2966 12
Yufeng Zhong, Long Xu, Jiebo Luo, Lin Ma: Contextual Modeling for 3D Dense Captioning on Point Clouds.
Scan2Cap		0.0849 18	0.0576 18	0.1073 18	0.0492 18	0.0970 18	0.2481 18
Dave Zhenyu Chen, Ali Gholami, Matthias Nießner and Angel X. Chang: Scan2Cap: Context-aware Dense Captioning in RGB-D Scans. CVPR 2021
X-Trans2Cap		0.1274 14	0.0808 15	0.1392 15	0.0653 15	0.1244 12	0.2795 16
Yuan, Zhihao and Yan, Xu and Liao, Yinghong and Guo, Yao and Li, Guanbin and Cui, Shuguang and Li, Zhen: X-Trans2Cap: Cross-Modal Knowledge Transfer Using Transformer for 3D Dense Captioning. CVPR 2022
SpaCap3D		0.1359 13	0.0883 13	0.1591 13	0.0738 13	0.1182 14	0.3275 9
Heng Wang, Chaoyi Zhang, Jianhui Yu, Weidong Cai: Spatiality-guided Transformer for 3D Dense Captioning on Point Clouds. the 31st International Joint Conference on Artificial Intelligence (IJCAI), 2022
MORE-xyz		0.1239 16	0.0796 16	0.1362 16	0.0631 16	0.1116 17	0.2648 17
Yang Jiao, Shaoxiang Chen, Zequn Jie, Jingjing Chen, Lin Ma, Yu-Gang Jiang: MORE: Multi_ORder RElation Mining for Dense Captioning in 3D Scenes. ECCV 2022
REMAN		0.1662 11	0.1070 11	0.1790 11	0.0815 11	0.1235 13	0.2927 14

3DJCG(Captioning)		0.1918 10	0.1350 8	0.2207 9	0.1013 9	0.1506 6	0.3867 8
Daigang Cai, Lichen Zhao, Jing Zhang†, Lu Sheng, Dong Xu: 3DJCG: A Unified Framework for Joint Dense Captioning and Visual Grounding on 3D Point Clouds. CVPR2022 Oral
SUN+		0.1148 17	0.0846 14	0.1564 14	0.0711 14	0.1143 16	0.2958 13

Chat-Scene-thres0.01		0.2053 9	0.1103 10	0.1884 10	0.0907 10	0.1527 5	0.5076 2

NOAH		0.1382 12	0.0901 12	0.1598 12	0.0747 12	0.1359 11	0.2977 11

Forest-xyz		0.2266 7	0.1363 7	0.2250 6	0.1027 7	0.1161 15	0.2825 15

CFM		0.2360 5	0.1417 5	0.2253 5	0.1034 5	0.1379 10	0.3008 10

vote2cap-detr		0.3128 3	0.1778 2	0.2842 4	0.1316 4	0.1825 2	0.4454 6
Sijin Chen, Hongyuan Zhu, Xin Chen, Yinjie Lei, Tao Chen, Gang YU, Taihao Li: End-to-End 3D Dense Captioning with Vote2Cap-DETR. CVPR 2023
Vote2Cap-DETR++		0.3360 1	0.1908 1	0.3012 1	0.1386 1	0.1864 1	0.5090 1
Sijin Chen, Hongyuan Zhu, Mingsheng Li, Xin Chen, Peng Guo, Yinjie Lei, Gang Yu, Taihao Li, Tao Chen: Vote2Cap-DETR++: Decoupling Localization and Describing for End-to-End 3D Dense Captioning.
TMP		0.3029 4	0.1728 3	0.2898 2	0.1332 3	0.1801 3	0.4605 5

Chat-Scene-all		0.1257 15	0.0671 17	0.1150 17	0.0554 17	0.1539 4	0.5076 2

D3Net - Speaker		0.2088 8	0.1335 9	0.2237 8	0.1022 8	0.1481 7	0.4198 7
Dave Zhenyu Chen, Qirui Wu, Matthias Niessner, Angel X. Chang: D3Net: A Unified Speaker-Listener Architecture for 3D Dense Captioning and Visual Grounding. 17th European Conference on Computer Vision (ECCV), 2022

ScanRefer Benchmark

Scan2Cap Benchmark