1. Home
  2. Benchmarks
  3. Object Detection on COCO minival
Dataset Object Detection on COCO minival

Object Detection on COCO minival

#
MODEL
REPOSITORY
BOX AP AP50 AP75 APS APM APL SPEED
PAPER
ε-REPRODUCES PAPER
1
HTC
(ResNeXt-101 64x4d-FPN, DCN, m…)
open-mmlab / mmdetection
0.506 0.701 0.551 0.328 0.547 0.664 3.2
--
2
GCNet
(ResNeXt-101 + DCN + cascade +…)
xvjiarui / GCNet
0.479 0.668 0.522 0.292 0.522 0.631 5.5
3
GCNet
(ResNeXt-101 + DCN + cascade +…)
xvjiarui / GCNet
0.478 0.668 0.518 0.281 0.523 0.634 5.4
4
Mask R-CNN
(ResNet-101-FPN, GN, Cascade)
tensorpack / tensorpack
0.477 0.661 0.521 0.304 0.507 0.615 2.4
5
Mask R-CNN
(ResNeXt-101 + DCN + cascade)
xvjiarui / GCNet
0.470 0.660 0.513 0.277 0.514 0.625 6.0
6
HTC
(ResNeXt-101 64x4d-FPN)
open-mmlab / mmdetection
0.469 0.660 0.512 0.280 0.507 0.621 3.5
--
7
RPDet
(ResNeXt-101-DCN, multi-scale)
microsoft / RepPoints
0.468 0.676 0.508 0.317 0.508 0.589 0.6
8
Mask R-CNN
(ResNeXt-101-FPN, syncBN, casc…)
xvjiarui / GCNet
0.464 0.653 0.506 0.281 0.507 0.605 6.3
9
RPDet
(ResNet-101-DCN, multi-scale)
microsoft / RepPoints
0.464 0.669 0.512 0.313 0.506 0.589 0.8
10
HTC
(ResNeXt-101 32x4d-FPN)
open-mmlab / mmdetection
0.461 0.651 0.502 0.275 0.498 0.612 3.9
--
11
Mask R-CNN
(ResNeXt-101-FPN, syncBN, casc…)
xvjiarui / GCNet
0.458 0.647 0.500 0.270 0.500 0.601 6.3
12
Cascade Mask R-CNN
(ResNeXt-101 64x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.457 0.641 0.500 0.262 0.496 0.600 5.7
13
HTC
(HRNetV2p-W32)
open-mmlab / mmdetection
0.453 0.636 0.491 0.270 0.484 0.595 3.8
--
14
Cascade Mask R-CNN
(ResNeXt-101 64x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.452 0.639 0.492 0.262 0.496 0.596 5.6
--
15
HTC
(ResNet-101-FPN)
open-mmlab / mmdetection
0.449 0.638 0.487 0.264 0.483 0.599 4.2
--
16
Cascade Mask R-CNN
(ResNeXt-101 32x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.447 0.630 0.489 0.259 0.487 0.589 6.9
--
17
Cascade R-CNN
(ResNeXt-101 64x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.447 0.631 0.490 0.258 0.483 0.588 7.0
--
18
Mask R-CNN
(ResNeXt-101-FPN, syncBN, casc…)
xvjiarui / GCNet
0.446 0.629 0.485 0.264 0.489 0.585 6.7
19
Mask R-CNN
(HRNetV2p-W32, cascade)
open-mmlab / mmdetection
0.445 0.623 0.486 0.261 0.479 0.585 6.5
--
20
RPDet
(ResNeXt-101-DCN)
microsoft / RepPoints
0.445 0.656 0.483 0.262 0.490 0.588 7.6
21
Cascade R-CNN
(ResNeXt-101 64x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.445 0.633 0.486 0.261 0.481 0.591 7.0
--
22
Cascade Mask R-CNN
(ResNeXt-101 32x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.444 0.626 0.486 0.254 0.481 0.587 6.8
--
23
RPDet
(ResNet-101-FPN, LR 2x, mstrain)
microsoft / RepPoints
0.441 0.646 0.483 0.299 0.479 0.554 0.9
--
24
Cascade R-CNN
(ResNeXt-101 32x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.440 0.625 0.480 0.253 0.478 0.581 9.2
--
25
Cascade R-CNN
(ResNeXt-101 32x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.436 0.622 0.474 0.250 0.477 0.574 9.1
--
26
Cascade R-CNN
(HRNetV2p-W32)
open-mmlab / mmdetection
0.436 0.615 0.476 0.255 0.464 0.573 8.7
--
27
Cascade Mask R-CNN
(ResNet-101-FPN, 20e LR)
open-mmlab / mmdetection
0.433 0.613 0.470 0.244 0.469 0.580 7.7
28
Mask R-CNN
(ResNeXt-101 32x4d-FPN, GN+WS,…)
open-mmlab / mmdetection
0.433 0.641 0.477 0.252 0.473 0.569 5.9
--
29
HTC
(ResNet-50-FPN, 20e LR)
open-mmlab / mmdetection
0.432 0.621 0.468 0.249 0.464 0.578 4.3
--
30
Mask R-CNN
(ResNeXt-101-FPN, syncBN, GC r…)
xvjiarui / GCNet
0.429 0.651 0.469 0.265 0.472 0.554 7.3
31
Libra R-CNN
(Faster R-CNN, ResNeXt-101 64x…)
open-mmlab / mmdetection
0.427 0.638 0.468 0.258 0.466 0.554
--
32
Cascade Mask R-CNN
(ResNet-101-FPN, 1x LR)
open-mmlab / mmdetection
0.426 0.607 0.467 0.238 0.464 0.569 7.5
--
33
Cascade R-CNN
(ResNet-101-FPN+, cascade)
open-mmlab / mmdetection
0.425 0.607 0.463 0.237 0.461 0.569 10.7
--
34
Mask R-CNN
(ResNeXt-101-FPN, syncBN, GC r…)
xvjiarui / GCNet
0.424 0.645 0.464 0.264 0.468 0.549 7.4
35
Cascade Mask R-CNN
(ResNet-50-FPN, 20e LR)
open-mmlab / mmdetection
0.423 0.605 0.460 0.237 0.457 0.564 8.4
--
36
RPDet
(ResNet-50, multi-scale train)
microsoft / RepPoints
0.422 0.626 0.462 0.293 0.457 0.528 1.2
37
Mask Scoring R-CNN
(ResNeXt-101 64x4d-FPN, LR 2x)
open-mmlab / mmdetection
0.422 0.634 0.461 0.242 0.460 0.561
--
38
HTC
(ResNet-50-FPN, 1x LR)
open-mmlab / mmdetection
0.421 0.608 0.459 0.239 0.455 0.562 3.6
--
39
Mask Scoring R-CNN
(ResNeXt-101 64x4d-FPN, LR 1x)
open-mmlab / mmdetection
0.420 0.640 0.460 0.248 0.465 0.543
--
40
Cascade R-CNN
(ResNet-101-FPN, 1x LR)
open-mmlab / mmdetection
0.420 0.603 0.459 0.232 0.459 0.563 10.6
--
41
FoveaBox
(ResNet-101, align, LR 2x, ms-…)
taokong / FoveaBox
0.419 0.625 0.452 0.251 0.464 0.543 10.4
--
42
Mask R-CNN
(HRNetV2p-W32)
open-mmlab / mmdetection
0.417 0.624 0.457 0.251 0.453 0.543 7.6
--
43
Mask R-CNN
(ResNet-101-FPN, syncBN, GC r4…)
xvjiarui / GCNet
0.417 0.637 0.455 0.247 0.462 0.539 8.1
44
RPDet
(ResNet-101-FPN-DCN, LR 2x)
microsoft / RepPoints
0.416 0.620 0.453 0.245 0.463 0.541 9.0
--
45
Mask R-CNN
(ResNeXt-101-FPN, syncBN, 1x LR)
xvjiarui / GCNet
0.415 0.632 0.452 0.243 0.461 0.539 7.9
46
Mask R-CNN
(ResNet-101-FPN, syncBN, GC r1…)
xvjiarui / GCNet
0.414 0.636 0.450 0.250 0.459 0.540 8.2
47
Faster R-CNN
(ResNeXt-101 64x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.413 0.633 0.452 0.244 0.458 0.534 7.5
--
48
Cascade Mask R-CNN
(ResNet-50-FPN, 1x LR)
open-mmlab / mmdetection
0.412 0.591 0.451 0.233 0.445 0.545 8.1
49
Cascade R-CNN
(ResNet-50-FPN, 20e LR)
open-mmlab / mmdetection
0.411 0.591 0.448 0.225 0.444 0.549 12.3
--
50
Mask R-CNN
(ResNet-50-FPN, GN+WS, LR 20-2…)
open-mmlab / mmdetection
0.410 0.619 0.448 0.245 0.446 0.537 7.9
--
51
Mask R-CNN
(HRNetV2p-W32, 1x LR)
open-mmlab / mmdetection
0.407 0.618 0.447 0.252 0.444 0.518 7.3
52
Mask Scoring R-CNN
(ResNet-101-FPN, LR 2x)
open-mmlab / mmdetection
0.407 0.617 0.446 0.236 0.446 0.536
--
53
Faster R-CNN
(ResNeXt-101 64x4d-FPN, 2x LR)
open-mmlab / mmdetection
0.407 0.620 0.446 0.229 0.445 0.536 7.6
--
54
Faster R-CNN
(HRNetV2p-W32)
open-mmlab / mmdetection
0.406 0.619 0.441 0.239 0.441 0.524 9.6
--
55
CRPN Faster R-CNN
(ResNet-50-FPN)
thangvubk / Cascade-RPN
0.405 0.586 0.442 0.219 0.434 0.545 9.3
--
56
Faster R-CNN
(ResNeXt-101 32x4d-FPN, 2x LR)
open-mmlab / mmdetection
0.404 0.619 0.441 0.233 0.446 0.529 10.2
--
57
Cascade R-CNN
(ResNet-50-FPN+)
open-mmlab / mmdetection
0.404 0.585 0.439 0.215 0.437 0.538 12.1
--
58
Mask R-CNN
(ResNet-50-FPN, GroupNorm)
tensorpack / tensorpack
0.404 0.609 0.440 0.240 0.433 0.538 2.6
59
RPDet
(ResNet-101)
microsoft / RepPoints
0.403 0.610 0.435 0.232 0.446 0.524 9.9
60
Libra R-CNN
(Faster R-CNN, ResNet-101-FPN)
open-mmlab / mmdetection
0.403 0.612 0.439 0.233 0.443 0.522
--
61
Mask R-CNN
(ResNet-50-FPN, GN+WS, LR 2x)
open-mmlab / mmdetection
0.402 0.608 0.435 0.235 0.433 0.534 8.2
--
62
Faster R-CNN
(ResNeXt-101 32x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.401 0.620 0.438 0.234 0.446 0.517 10.1
--
63
FoveaBox
(ResNet-50, align, LR 2x, ms-t…)
taokong / FoveaBox
0.401 0.608 0.425 0.238 0.440 0.518 12.0
--
64
Mask R-CNN
(ResNet-50-FPN, GC r4, 2x LR)
xvjiarui / GCNet
0.400 0.621 0.431 0.244 0.440 0.517
65
RetinaNet
(ResNeXt-101 64x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.400 0.609 0.430 0.235 0.444 0.526 6.9
--
66
Mask Scoring R-CNN
(ResNet-101-FPN, LR 1x)
open-mmlab / mmdetection
0.398 0.612 0.434 0.233 0.439 0.514
--
67
Mask R-CNN
(ResNet-50-FPN, GC r16, 2x LR)
xvjiarui / GCNet
0.397 0.617 0.431 0.238 0.436 0.512
68
Mask R-CNN
(ResNet-101-FPN, syncBN, 1x LR)
xvjiarui / GCNet
0.397 0.612 0.428 0.228 0.443 0.510
69
Mask R-CNN
(ResNet-101-FPN, fixBN, 1x LR)
xvjiarui / GCNet
0.397 0.612 0.428 0.228 0.443 0.510
--
70
RetinaNet
(ResNeXt-101 64x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.396 0.603 0.423 0.216 0.435 0.535 7.0
--
71
Faster R-CNN
(HRNetV2p-W32, 1x LR)
open-mmlab / mmdetection
0.395 0.610 0.431 0.236 0.429 0.510 9.2
72
FoveaBox
(ResNet-101, align, LR 2x)
taokong / FoveaBox
0.394 0.599 0.421 0.218 0.437 0.521 10.6
--
73
Faster R-CNN
(ResNet-101-FPN, 2x LR)
open-mmlab / mmdetection
0.394 0.606 0.430 0.221 0.436 0.521 11.9
--
74
RetinaNet
(ResNeXt-101 32x4d-FPN, 20e LR)
open-mmlab / mmdetection
0.393 0.598 0.423 0.210 0.436 0.523 9.1
--
75
Mask R-CNN
(HRNetV2p-W18)
open-mmlab / mmdetection
0.392 0.601 0.429 0.241 0.422 0.507 7.2
--
76
RetinaNet
(ResNeXt-101 32x4d-FPN, 1x LR)
open-mmlab / mmdetection
0.390 0.594 0.417 0.226 0.434 0.509 8.9
--
77
Mask R-CNN
(ResNet-50-FPN, 2x)
tensorpack / tensorpack
0.389 0.597 0.421 0.219 0.426 0.511 2.8
--
78
Faster R-CNN
(ResNet-50-FPN, GN+WS, LR 1x)
open-mmlab / mmdetection
0.389 0.604 0.421 0.235 0.424 0.504 9.9
--
79
RPDet
(ResNet-50)
microsoft / RepPoints
0.386 0.595 0.412 0.221 0.426 0.506 11.4
80
FoveaBox
(ResNet-101-FPN, 800x800)
taokong / FoveaBox
0.385 0.581 0.409 0.213 0.426 0.508 11.7
81
Libra R-CNN
(Faster R-CNN, ResNet-50-FPN)
open-mmlab / mmdetection
0.385 0.595 0.425 0.229 0.418 0.489
--
82
FoveaBox
(ResNet-101, LR 1x)
taokong / FoveaBox
0.385 0.581 0.415 0.215 0.428 0.503 11.5
--
83
Faster R-CNN
(ResNet-101-FPN, 1x LR)
open-mmlab / mmdetection
0.385 0.603 0.416 0.223 0.430 0.498 11.8
--
84
Faster R-CNN
(HRNetV2p-W18)
open-mmlab / mmdetection