Browse › Computer Vision › Instance Segmentation › COCO minival

Instance Segmentation archive 2025-07-28

COCO minival Benchmark (Instance Segmentation)

93 rows 89 with code listed 9 metrics Dataset page

Instance Segmentation is a computer vision task that involves identifying and separating individual objects within an image, including detecting the boundaries of each object and assigning a unique label to each object. The goal of instance segmentation is to produce a pixel-wise segmentation map of the image, where each pixel is assigned to a specific object instance.

The archive carries no text for this table; the description above is the archive's text for the task Instance Segmentation. archive 2025-07-28

Over time archive 2025-07-28

The chart needs JavaScript; the table below carries every value.

Direction inferred from the metric name, not from the archive: mask AP (higher is better), Params (M) (lower is better), box AP (higher is better). Not inferred (points only, no best-so-far line): AP50, AP75, APL, APM, APS, GFLOPs. Points are placed at the row's paper date; 93 of 93 rows carry one.

Results archive 2025-07-28

Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.

Paper Code Ran Syntology Report
1 Co-DETR 56.679.762.874.659.738.9 ✓ Paper Code 2022 0 of 5 ran · 5 unverified report
2 ViT-CoMer-L (Mask RCNN, DINOv2) 55.9 – Paper Code 2024 linked, not harvested report
3 InternImage-H 55.480.161.574.458.437.9 ✓ Paper Code 2022 2 of 4 ran · 2 unverified report
4 EVA 55.079.460.972.058.437.6 ✓ Paper Code 2022 1 of 3 ran · 2 unverified report
5 Mask Frozen-DETR 54.978.960.872.958.4 37.2 ✓ Paper – 2023 no code linked report
6 MasK DINO (SwinL, multi-scale) 54.5 ✓ Paper Code 2022 11 of 13 ran · 2 unverified report
7 ViT-Adapter-L (HTC++, BEiTv2, O365, multi-scale) 54.2 ✓ Paper Code 2022 linked, not harvested report
8 GLEE-Pro 54.2 ✓ Paper Code 2023 8 of 13 ran · 5 unverified report
9 SwinV2-G (HTC++) 53.7 ✓ Paper Code 2021 3 of 30 ran · 27 unverified report
10 ViTDet, ViT-H Cascade (multiscale) 53.1 – Paper Code 2022 1 of 4 ran · 3 unverified report
11 GLEE-Plus 53.0 ✓ Paper Code 2023 8 of 13 ran · 5 unverified report
12 Mask DINO (SwinL) 52.6 – Paper Code 2022 11 of 13 ran · 2 unverified report
13 Soft Teacher + Swin-L(HTC++, multi-scale) 52.5 ✓ Paper Code 2021 linked, not harvested report
14 ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) 52.5 – Paper Code 2022 linked, not harvested report
15 ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) 52.2 – Paper Code 2022 linked, not harvested report
16 ViTDet, ViT-H Cascade 52 – Paper Code 2022 1 of 4 ran · 3 unverified report
17 Soft Teacher + Swin-L(HTC++, single-scale) 51.9 ✓ Paper Code 2021 linked, not harvested report
18 CBNetV2 (Dual-Swin-L HTC, multi-scale) 51.8 – Paper Code 2021 2 of 3 ran · 1 unverified report
19 Frozen Backbone, SwinV2-G-ext22K (HTC) 51.6 – Paper – 2022 no code linked report
20 CBNetV2 (Dual-Swin-L HTC, multi-scale) 51 – Paper Code 2021 2 of 3 ran · 1 unverified report
21 Focal-L (HTC++, multi-scale) 50.9 – Paper Code 2021 3 of 3 ran · 0 unverified report
22 DiNAT-L (single-scale, Mask2Former) 50.875.0 – Paper Code 2022 linked, not harvested report
23 MViTv2-L (Cascade Mask R-CNN, multi-scale, IN21k pre-train) 50.5 – Paper Code 2021 linked, not harvested report
24 Swin-L (HTC++, multi scale) 50.4 – Paper Code 2021 108 of 207 ran · 99 unverified report
25 MOAT-3 (IN-22K pretraining, single-scale) 50.3 – Paper Code 2022 linked, not harvested report
26 Mask2Former (Swin-L) 50.1 – Paper Code 2021 2 of 8 ran · 6 unverified report
27 Swin-L (HTC++, single scale) 49.5 – Paper Code 2021 108 of 207 ran · 99 unverified report
28 MOAT-2 (IN-22K pretraining, single-scale) 49.3 – Paper Code 2022 linked, not harvested report
29 MOAT-1 (IN-1K pretraining, single-scale) 49.0 – Paper Code 2022 linked, not harvested report
30 QueryInst (single scale) 48.974.053.968.352.630.8 – Paper Code 2021 1 of 3 ran · 2 unverified report
31 Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) 48.9 ✓ Paper Code 2020 2 of 2 ran · 0 unverified report
32 InternImage-XL 48.81782387 – Paper Code 2022 2 of 4 ran · 2 unverified report
33 CenterNet2 (Swin-L w/ X-Paste + Copy-Paste) 48.8 – Paper Code 2022 1 of 3 ran · 2 unverified report
34 Heira-L 48.6 – Paper Code 2023 0 of 6 ran · 6 unverified report
35 InternImage-L 48.5139927756.1 – Paper Code 2022 2 of 4 ran · 2 unverified report
36 MViTv2-H (Cascade Mask R-CNN, single-scale, IN21k pre-train) 48.5 – Paper Code 2021 linked, not harvested report
37 GLEE-Lite 48.4 ✓ Paper Code 2023 8 of 13 ran · 5 unverified report
38 MOAT-0 (IN-1K pretraining, single-scale) 47.4 – Paper Code 2022 linked, not harvested report
39 MViTv2-L (Cascade Mask R-CNN, single-scale) 47.1 – Paper Code 2021 linked, not harvested report
40 MPViT-B (Cascade Mask R-CNN, multi-scale, IN1k pre-train) 47.0 – Paper Code 2021 linked, not harvested report
41 tiny-MOAT-3 (IN-1K pretraining, single-scale) 47.0 – Paper Code 2022 linked, not harvested report
42 Cascade Eff-B7 NAS-FPN (1280) 46.8 – Paper Code 2020 2 of 2 ran · 0 unverified report
43 ResNeSt-200 (multi-scale) 46.25 – Paper Code 2020 8 of 48 ran · 40 unverified report
44 MViT-L (Mask R-CNN, single-scale) 46.2 – Paper Code 2021 linked, not harvested report
45 RetinaNet (SpineNet-190, 1536x1536) 46.1 – Paper Code 2019 linked, not harvested report
46 MPViT-B (Cascade R-CNN, sinlge-scale, IN-1K pre-train) 45.8 – Paper Code 2021 linked, not harvested report
47 Mask R-CNN (ViL Base, multi-scale, 3x lr) 45.749.9 – Paper Code 2021 6 of 10 ran · 4 unverified report
48 Mask R-CNN (ViL Base, 1x lr) 45.167.249.3 – Paper Code 2021 6 of 10 ran · 4 unverified report
49 tiny-MOAT-2 (IN-1K pretraining, single-scale) 45.0 – Paper Code 2022 linked, not harvested report
50 GCNet (ResNeXt-101 + DCN + cascade + GC r4) 44.767.948.4 – Paper Code 2020 linked, not harvested report
51 tiny-MOAT-1 (IN-1K pretraining, single-scale) 44.6 – Paper Code 2022 linked, not harvested report
52 InternImage-S 44.53406949.7 – Paper Code 2022 2 of 4 ran · 2 unverified report
53 ResNeSt-200-DCN (single-scale) 44.5 – Paper Code 2020 8 of 48 ran · 40 unverified report
54 ELSA-S (Cascade Mask RCNN) 44.467.847.8 – Paper Code 2021 1 of 1 ran · 0 unverified report
55 BoTNet 200 (Mask R-CNN, single scale, 72 epochs) 44.4 – Paper Code 2021 26 of 49 ran · 23 unverified report
56 DaViT-T (Mask R-CNN, 36 epochs) 44.3 – Paper Code 2022 8 of 15 ran · 7 unverified report
57 ResNeSt-200 (single-scale) 44.21 – Paper Code 2020 8 of 48 ran · 40 unverified report
58 InternImage-T 43.72704949.1 – Paper Code 2022 2 of 4 ran · 2 unverified report
59 BoTNet 152 (Mask R-CNN, single scale, 72 epochs) 43.7 – Paper Code 2021 26 of 49 ran · 23 unverified report
60 XCiT-M24/8 43.7 – Paper Code 2021 3 of 14 ran · 11 unverified report
61 tiny-MOAT-0 (IN-1K pretraining, single-scale) 43.3 – Paper Code 2022 linked, not harvested report
62 ELSA-S (Mask RCNN) 43.067.346.4 – Paper Code 2021 1 of 1 ran · 0 unverified report
63 XCiT-S24/8 43.0 – Paper Code 2021 3 of 14 ran · 11 unverified report
64 CenterMask-VoVNetV2-99 (multi-scale) 42.5 – Paper Code 2019 1 of 2 ran · 1 unverified report
65 ResNeSt-101 (single-scale) 41.56 – Paper Code 2020 8 of 48 ran · 40 unverified report
66 SIW 41.4 – Paper – 2022 no code linked report
67 Res2Net-101+HTC 41.3 – Paper Code 2019 3 of 9 ran · 6 unverified report
68 HTC (HRNetV2p-W48) 41.0 – Paper Code 2019 8 of 25 ran · 17 unverified report
69 HTC (HRNetV2p-W48) 41.0 – Paper Code 2019 3 of 34 ran · 31 unverified report
70 GCNet (ResNeXt-101 + DCN + cascade + GC r16) 40.9 – Paper Code 2019 linked, not harvested report
71 BoTNet 50 (72 epochs) 40.7 – Paper Code 2021 26 of 49 ran · 23 unverified report
72 R3-CNN (ResNet-50-FPN, DCN) 40.461.34456.143.622.3 – Paper Code 2021 linked, not harvested report
73 Mask R-CNN (ResNext-152, +1 NL) 40.3 – Paper Code 2017 3 of 4 ran · 1 unverified report
74 Mask R-CNN-FPN (AOGNet-40M) 40.263.243.3 – Paper Code 2019 linked, not harvested report
75 R3-CNN (ResNet-50-FPN, GC-Net) 40.261.143.542.822.6 – Paper Code 2021 linked, not harvested report
76 CenterMask-VoVNetV2-99-3x 40.2 – Paper Code 2019 1 of 2 ran · 1 unverified report
77 R3-CNN (ResNet-50-FPN, GRoIE) 39.158.842.354.342.120.7 – Paper Code 2021 linked, not harvested report
78 Mask Scoring R-CNN (ResNet-101-FPN-DCN) 39.1 – Paper Code 2019 1 of 3 ran · 2 unverified report
79 Mask R-CNN-FPN (ResNeXt-101, GN+WS) 38.3461.0740.8256.0841.7318.32 – Paper Code 2019 linked, not harvested report
80 R3-CNN (ResNet-50-FPN) 38.25841.452.84120.4 – Paper Code 2021 linked, not harvested report
81 HTC (ResNet-50) 38.2 – Paper Code 2019 0 of 11 ran · 11 unverified report
82 Mask Scoring R-CNN (ResNet-101 FPN) 38.2 – Paper Code 2019 1 of 3 ran · 2 unverified report
83 PANet (ResNet-50) 37.8 – Paper Code 2018 1 of 4 ran · 3 unverified report
84 GCnet (ResNet-50-FPN, GRoIE) 37.259.339.851.24120.2 – Paper Code 2020 linked, not harvested report
85 Mask R-CNN (FPN, X-volution, SA) 37.253.14019.2 – Paper – 2021 no code linked report
86 Mask R-CNN (ResNet-101, +1 NL) 37.1 – Paper Code 2017 3 of 4 ran · 1 unverified report
87 Mask Scoring R-CNN (ResNet-50 FPN) 36.0 – Paper Code 2019 1 of 3 ran · 2 unverified report
88 Mask R-CNN (ResNet-50-FPN, GRoIE) 35.857.138.048.73919.1 – Paper Code 2020 linked, not harvested report
89 Faster R-CNN (Res2Net-50) 35.657.653.737.915.7 – Paper Code 2019 3 of 9 ran · 6 unverified report
90 Mask R-CNN (ResNet-50, +1 NL) 35.5 – Paper Code 2017 3 of 4 ran · 1 unverified report
91 Mask R-CNN (ResNet-50, ACNet) 35.2 – Paper Code 2019 linked, not harvested report
92 YOLACT-550 (ResNet-50) 29.9 – Paper Code 2019 10 of 21 ran · 11 unverified report
93 InternImage-B 501115 – Paper Code 2022 2 of 4 ran · 2 unverified report

All 93 rows shown. 93 link to a paper page on this site; 13 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28

Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 56 rows have a graph line, from 30 distinct papers; 53 rows (27 papers) have at least one sample that ran. Counting each paper once: Syntology ran 227 of 557 samples; 330 unverified. Separately, 133 of those 557 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections