paper-with-me

Object Detection 벤치마크

Object Detection on COCO test-dev

1125개 결과 · ⬇ CSV · JSON

box mAP

35.9 43.42 50.95 58.48 66 2016-12 2026-09 Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + TDM — 36.8 (2016-12-20) Faster R-CNN + TDM — 36.8 (2016-12-20) Faster R-CNN + TDM — 36.8 (2016-12-20) Faster R-CNN + TDM — 36.8 (2016-12-20) Faster R-CNN + TDM — 36.8 (2016-12-20) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) Mask R-CNN (ResNet-101-FPN) — 38.2 (2017-03-20) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) Mask R-CNN (ResNet-101-FPN) — 38.2 (2017-03-20) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) Mask R-CNN (ResNet-101-FPN) — 38.2 (2017-03-20) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) Mask R-CNN (ResNet-101-FPN) — 38.2 (2017-03-20) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) Mask R-CNN (ResNet-101-FPN) — 38.2 (2017-03-20) Faster R-CNN (ImageNet+300M) — 37.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 37.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 37.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 37.4 (2017-07-10) Faster R-CNN (ImageNet+300M) — 37.4 (2017-07-10) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RetinaNet (ResNet-101-FPN) — 39.1 (2017-08-07) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RetinaNet (ResNet-101-FPN) — 39.1 (2017-08-07) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RetinaNet (ResNet-101-FPN) — 39.1 (2017-08-07) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RetinaNet (ResNet-101-FPN) — 39.1 (2017-08-07) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RetinaNet (ResNet-101-FPN) — 39.1 (2017-08-07) FPN (ResNet101 backbone) — 39.5 (2017-08-28) FPN (ResNet101 backbone) — 39.5 (2017-08-28) FPN (ResNet101 backbone) — 39.5 (2017-08-28) FPN (ResNet101 backbone) — 39.5 (2017-08-28) FPN (ResNet101 backbone) — 39.5 (2017-08-28) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) RefineDet512+ (VGG-16) — 37.6 (2017-11-18) RefineDet512 (ResNet-101) — 36.4 (2017-11-18) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) RefineDet512+ (VGG-16) — 37.6 (2017-11-18) RefineDet512 (ResNet-101) — 36.4 (2017-11-18) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) RefineDet512+ (VGG-16) — 37.6 (2017-11-18) RefineDet512 (ResNet-101) — 36.4 (2017-11-18) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) RefineDet512+ (VGG-16) — 37.6 (2017-11-18) RefineDet512 (ResNet-101) — 36.4 (2017-11-18) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) RefineDet512+ (VGG-16) — 37.6 (2017-11-18) RefineDet512 (ResNet-101) — 36.4 (2017-11-18) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) D-RFCN + SNIP (ResNet-101, multi-scale) — 43.4 (2017-11-22) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) D-RFCN + SNIP (ResNet-101, multi-scale) — 43.4 (2017-11-22) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) D-RFCN + SNIP (ResNet-101, multi-scale) — 43.4 (2017-11-22) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) D-RFCN + SNIP (ResNet-101, multi-scale) — 43.4 (2017-11-22) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) D-RFCN + SNIP (ResNet-101, multi-scale) — 43.4 (2017-11-22) Cascade R-CNN (ResNet-101-FPN+, cascade) — 42.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+, cascade) — 40.6 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+) — 38.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+) — 36.5 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+, cascade) — 42.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+, cascade) — 40.6 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+) — 38.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+) — 36.5 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+, cascade) — 42.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+, cascade) — 40.6 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+) — 38.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+) — 36.5 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+, cascade) — 42.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+, cascade) — 40.6 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+) — 38.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+) — 36.5 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+, cascade) — 42.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+, cascade) — 40.6 (2017-12-03) Cascade R-CNN (ResNet-101-FPN+) — 38.8 (2017-12-03) Cascade R-CNN (ResNet-50-FPN+) — 36.5 (2017-12-03) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) SNIPER (ResNet-101) — 46.1 (2018-05-23) SNIPER (ResNet-50) — 43.5 (2018-05-23) SNIPER (ResNet-101) — 46.1 (2018-05-23) SNIPER (ResNet-50) — 43.5 (2018-05-23) SNIPER (ResNet-101) — 46.1 (2018-05-23) SNIPER (ResNet-50) — 43.5 (2018-05-23) SNIPER (ResNet-101) — 46.1 (2018-05-23) SNIPER (ResNet-50) — 43.5 (2018-05-23) SNIPER (ResNet-101) — 46.1 (2018-05-23) SNIPER (ResNet-50) — 43.5 (2018-05-23) IoU-Net — 40.6 (2018-07-30) IoU-Net — 40.6 (2018-07-30) IoU-Net — 40.6 (2018-07-30) IoU-Net — 40.6 (2018-07-30) IoU-Net — 40.6 (2018-07-30) CornerNet511 (Hourglass-104, multi-scale) — 42.1 (2018-08-03) CornerNet511 (Hourglass-52, single-scale) — 37.8 (2018-08-03) CornerNet511 (Hourglass-104, multi-scale) — 42.1 (2018-08-03) CornerNet511 (Hourglass-52, single-scale) — 37.8 (2018-08-03) CornerNet511 (Hourglass-104, multi-scale) — 42.1 (2018-08-03) CornerNet511 (Hourglass-52, single-scale) — 37.8 (2018-08-03) CornerNet511 (Hourglass-104, multi-scale) — 42.1 (2018-08-03) CornerNet511 (Hourglass-52, single-scale) — 37.8 (2018-08-03) CornerNet511 (Hourglass-104, multi-scale) — 42.1 (2018-08-03) CornerNet511 (Hourglass-52, single-scale) — 37.8 (2018-08-03) ResNet-50-FPN Mask R-CNN + KL Loss + var voting + soft-NMS — 40.4 (2018-09-23) ResNet-50-FPN Mask R-CNN + KL Loss + var voting + soft-NMS — 40.4 (2018-09-23) ResNet-50-FPN Mask R-CNN + KL Loss + var voting + soft-NMS — 40.4 (2018-09-23) ResNet-50-FPN Mask R-CNN + KL Loss + var voting + soft-NMS — 40.4 (2018-09-23) ResNet-50-FPN Mask R-CNN + KL Loss + var voting + soft-NMS — 40.4 (2018-09-23) M2Det (VGG-16, multi-scale) — 44.2 (2018-11-12) M2Det (ResNet-101, multi-scale) — 43.9 (2018-11-12) M2Det (VGG-16, single-scale) — 41.0 (2018-11-12) M2Det (ResNet-101, single-scale) — 38.8 (2018-11-12) M2Det (VGG-16, multi-scale) — 44.2 (2018-11-12) M2Det (ResNet-101, multi-scale) — 43.9 (2018-11-12) M2Det (VGG-16, single-scale) — 41.0 (2018-11-12) M2Det (ResNet-101, single-scale) — 38.8 (2018-11-12) M2Det (VGG-16, multi-scale) — 44.2 (2018-11-12) M2Det (ResNet-101, multi-scale) — 43.9 (2018-11-12) M2Det (VGG-16, single-scale) — 41.0 (2018-11-12) M2Det (ResNet-101, single-scale) — 38.8 (2018-11-12) M2Det (VGG-16, multi-scale) — 44.2 (2018-11-12) M2Det (ResNet-101, multi-scale) — 43.9 (2018-11-12) M2Det (VGG-16, single-scale) — 41.0 (2018-11-12) M2Det (ResNet-101, single-scale) — 38.8 (2018-11-12) M2Det (VGG-16, multi-scale) — 44.2 (2018-11-12) M2Det (ResNet-101, multi-scale) — 43.9 (2018-11-12) M2Det (VGG-16, single-scale) — 41.0 (2018-11-12) M2Det (ResNet-101, single-scale) — 38.8 (2018-11-12) GHM-C + GHM-R (RetinaNet-FPN-ResNeXt-101) — 41.6 (2018-11-13) GHM-C + GHM-R (RetinaNet-FPN-ResNeXt-101) — 41.6 (2018-11-13) GHM-C + GHM-R (RetinaNet-FPN-ResNeXt-101) — 41.6 (2018-11-13) GHM-C + GHM-R (RetinaNet-FPN-ResNeXt-101) — 41.6 (2018-11-13) GHM-C + GHM-R (RetinaNet-FPN-ResNeXt-101) — 41.6 (2018-11-13) DCNv2 (ResNet-101, multi-scale) — 46.0 (2018-11-27) DCNv2 (ResNet-101, multi-scale) — 46.0 (2018-11-27) DCNv2 (ResNet-101, multi-scale) — 46.0 (2018-11-27) DCNv2 (ResNet-101, multi-scale) — 46.0 (2018-11-27) DCNv2 (ResNet-101, multi-scale) — 46.0 (2018-11-27) Grid R-CNN (ResNeXt-101-FPN) — 43.2 (2018-11-29) Grid R-CNN (ResNeXt-101-FPN) — 43.2 (2018-11-29) Grid R-CNN (ResNeXt-101-FPN) — 43.2 (2018-11-29) Grid R-CNN (ResNeXt-101-FPN) — 43.2 (2018-11-29) Grid R-CNN (ResNeXt-101-FPN) — 43.2 (2018-11-29) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) TridentNet (ResNet-101) — 42.7 (2019-01-07) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) TridentNet (ResNet-101) — 42.7 (2019-01-07) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) TridentNet (ResNet-101) — 42.7 (2019-01-07) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) TridentNet (ResNet-101) — 42.7 (2019-01-07) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) TridentNet (ResNet-101) — 42.7 (2019-01-07) RetinaMask (ResNeXt-101-FPN-GN) — 42.6 (2019-01-10) GA-Faster-RCNN — 39.8 (2019-01-10) RetinaMask (ResNet-50-FPN) — 39.4 (2019-01-10) RetinaMask (ResNeXt-101-FPN-GN) — 42.6 (2019-01-10) GA-Faster-RCNN — 39.8 (2019-01-10) RetinaMask (ResNet-50-FPN) — 39.4 (2019-01-10) RetinaMask (ResNeXt-101-FPN-GN) — 42.6 (2019-01-10) GA-Faster-RCNN — 39.8 (2019-01-10) RetinaMask (ResNet-50-FPN) — 39.4 (2019-01-10) RetinaMask (ResNeXt-101-FPN-GN) — 42.6 (2019-01-10) GA-Faster-RCNN — 39.8 (2019-01-10) RetinaMask (ResNet-50-FPN) — 39.4 (2019-01-10) RetinaMask (ResNeXt-101-FPN-GN) — 42.6 (2019-01-10) GA-Faster-RCNN — 39.8 (2019-01-10) RetinaMask (ResNet-50-FPN) — 39.4 (2019-01-10) HTC (ResNeXt-101-FPN) — 47.1 (2019-01-22) HTC (ResNeXt-101-FPN) — 47.1 (2019-01-22) HTC (ResNeXt-101-FPN) — 47.1 (2019-01-22) HTC (ResNeXt-101-FPN) — 47.1 (2019-01-22) HTC (ResNeXt-101-FPN) — 47.1 (2019-01-22) ExtremeNet (Hourglass-104, multi-scale) — 43.7 (2019-01-23) ExtremeNet (Hourglass-104, single-scale) — 40.2 (2019-01-23) ExtremeNet (Hourglass-104, multi-scale) — 43.7 (2019-01-23) ExtremeNet (Hourglass-104, single-scale) — 40.2 (2019-01-23) ExtremeNet (Hourglass-104, multi-scale) — 43.7 (2019-01-23) ExtremeNet (Hourglass-104, single-scale) — 40.2 (2019-01-23) ExtremeNet (Hourglass-104, multi-scale) — 43.7 (2019-01-23) ExtremeNet (Hourglass-104, single-scale) — 40.2 (2019-01-23) ExtremeNet (Hourglass-104, multi-scale) — 43.7 (2019-01-23) ExtremeNet (Hourglass-104, single-scale) — 40.2 (2019-01-23) FSAF (ResNeXt-101, multi-scale) — 44.6 (2019-03-02) FSAF (ResNet-101, single-scale) — 40.9 (2019-03-02) FSAF (ResNeXt-101, multi-scale) — 44.6 (2019-03-02) FSAF (ResNet-101, single-scale) — 40.9 (2019-03-02) FSAF (ResNeXt-101, multi-scale) — 44.6 (2019-03-02) FSAF (ResNet-101, single-scale) — 40.9 (2019-03-02) FSAF (ResNeXt-101, multi-scale) — 44.6 (2019-03-02) FSAF (ResNet-101, single-scale) — 40.9 (2019-03-02) FSAF (ResNeXt-101, multi-scale) — 44.6 (2019-03-02) FSAF (ResNet-101, single-scale) — 40.9 (2019-03-02) InterNet (ResNet-101-FPN, multi-scale) — 44.2 (2019-03-28) InterNet (ResNet-101-FPN, multi-scale) — 44.2 (2019-03-28) InterNet (ResNet-101-FPN, multi-scale) — 44.2 (2019-03-28) InterNet (ResNet-101-FPN, multi-scale) — 44.2 (2019-03-28) InterNet (ResNet-101-FPN, multi-scale) — 44.2 (2019-03-28) FCOS (ResNeXt-64x4d-101-FPN 4 + improvements) — 44.7 (2019-04-02) FCOS (ResNeXt-101-64x4d-FPN) — 43.2 (2019-04-02) FCOS (ResNeXt-32x8d-101-FPN) — 42.7 (2019-04-02) FCOS (HRNet-W32-5l) — 42.0 (2019-04-02) FCOS (ResNeXt-64x4d-101-FPN 4 + improvements) — 44.7 (2019-04-02) FCOS (ResNeXt-101-64x4d-FPN) — 43.2 (2019-04-02) FCOS (ResNeXt-32x8d-101-FPN) — 42.7 (2019-04-02) FCOS (HRNet-W32-5l) — 42.0 (2019-04-02) FCOS (ResNeXt-64x4d-101-FPN 4 + improvements) — 44.7 (2019-04-02) FCOS (ResNeXt-101-64x4d-FPN) — 43.2 (2019-04-02) FCOS (ResNeXt-32x8d-101-FPN) — 42.7 (2019-04-02) FCOS (HRNet-W32-5l) — 42.0 (2019-04-02) FCOS (ResNeXt-64x4d-101-FPN 4 + improvements) — 44.7 (2019-04-02) FCOS (ResNeXt-101-64x4d-FPN) — 43.2 (2019-04-02) FCOS (ResNeXt-32x8d-101-FPN) — 42.7 (2019-04-02) FCOS (HRNet-W32-5l) — 42.0 (2019-04-02) FCOS (ResNeXt-64x4d-101-FPN 4 + improvements) — 44.7 (2019-04-02) FCOS (ResNeXt-101-64x4d-FPN) — 43.2 (2019-04-02) FCOS (ResNeXt-32x8d-101-FPN) — 42.7 (2019-04-02) FCOS (HRNet-W32-5l) — 42.0 (2019-04-02) Libra R-CNN (ResNeXt-101-FPN) — 43.0 (2019-04-04) Libra R-CNN (ResNeXt-101-FPN) — 43.0 (2019-04-04) Libra R-CNN (ResNeXt-101-FPN) — 43.0 (2019-04-04) Libra R-CNN (ResNeXt-101-FPN) — 43.0 (2019-04-04) Libra R-CNN (ResNeXt-101-FPN) — 43.0 (2019-04-04) FoveaBox (ResNeXt-101) — 43.9 (2019-04-08) FoveaBox (ResNeXt-101) — 42.1 (2019-04-08) FoveaBox (ResNeXt-101) — 41.9 (2019-04-08) FoveaBox (ResNeXt-101) — 43.9 (2019-04-08) FoveaBox (ResNeXt-101) — 42.1 (2019-04-08) FoveaBox (ResNeXt-101) — 41.9 (2019-04-08) FoveaBox (ResNeXt-101) — 43.9 (2019-04-08) FoveaBox (ResNeXt-101) — 42.1 (2019-04-08) FoveaBox (ResNeXt-101) — 41.9 (2019-04-08) FoveaBox (ResNeXt-101) — 43.9 (2019-04-08) FoveaBox (ResNeXt-101) — 42.1 (2019-04-08) FoveaBox (ResNeXt-101) — 41.9 (2019-04-08) FoveaBox (ResNeXt-101) — 43.9 (2019-04-08) FoveaBox (ResNeXt-101) — 42.1 (2019-04-08) FoveaBox (ResNeXt-101) — 41.9 (2019-04-08) CenterNet-DLA (DLA-34, multi-scale) — 41.6 (2019-04-16) CenterNet-DLA (DLA-34, multi-scale) — 41.6 (2019-04-16) CenterNet-DLA (DLA-34, multi-scale) — 41.6 (2019-04-16) CenterNet-DLA (DLA-34, multi-scale) — 41.6 (2019-04-16) CenterNet-DLA (DLA-34, multi-scale) — 41.6 (2019-04-16) CenterNet511 (Hourglass-104, multi-scale) — 47.0 (2019-04-17) CenterNet511 (Hourglass-104, multi-scale) — 47.0 (2019-04-17) CenterNet511 (Hourglass-104, multi-scale) — 47.0 (2019-04-17) CenterNet511 (Hourglass-104, multi-scale) — 47.0 (2019-04-17) CenterNet511 (Hourglass-104, multi-scale) — 47.0 (2019-04-17) CornerNet-Saccade (Hourglass-104, multi-scale) — 43.2 (2019-04-18) CornerNet-Saccade (Hourglass-104, multi-scale) — 43.2 (2019-04-18) CornerNet-Saccade (Hourglass-104, multi-scale) — 43.2 (2019-04-18) CornerNet-Saccade (Hourglass-104, multi-scale) — 43.2 (2019-04-18) CornerNet-Saccade (Hourglass-104, multi-scale) — 43.2 (2019-04-18) AA-ResNet-10 + RetinaNet — 39.2 (2019-04-22) AA-ResNet-10 + RetinaNet — 39.2 (2019-04-22) AA-ResNet-10 + RetinaNet — 39.2 (2019-04-22) AA-ResNet-10 + RetinaNet — 39.2 (2019-04-22) AA-ResNet-10 + RetinaNet — 39.2 (2019-04-22) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 48.4 (2019-04-25) RPDet (ResNet-101-DCN, multi-scale) — 46.5 (2019-04-25) RPDet (ResNet-101-DCN) — 42.8 (2019-04-25) RPDet (ResNet-101) — 41.0 (2019-04-25) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 48.4 (2019-04-25) RPDet (ResNet-101-DCN, multi-scale) — 46.5 (2019-04-25) RPDet (ResNet-101-DCN) — 42.8 (2019-04-25) RPDet (ResNet-101) — 41.0 (2019-04-25) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 48.4 (2019-04-25) RPDet (ResNet-101-DCN, multi-scale) — 46.5 (2019-04-25) RPDet (ResNet-101-DCN) — 42.8 (2019-04-25) RPDet (ResNet-101) — 41.0 (2019-04-25) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 48.4 (2019-04-25) RPDet (ResNet-101-DCN, multi-scale) — 46.5 (2019-04-25) RPDet (ResNet-101-DCN) — 42.8 (2019-04-25) RPDet (ResNet-101) — 41.0 (2019-04-25) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 48.4 (2019-04-25) RPDet (ResNet-101-DCN, multi-scale) — 46.5 (2019-04-25) RPDet (ResNet-101-DCN) — 42.8 (2019-04-25) RPDet (ResNet-101) — 41.0 (2019-04-25) WSMA-Seg — 38.1 (2019-04-30) WSMA-Seg — 38.1 (2019-04-30) WSMA-Seg — 38.1 (2019-04-30) WSMA-Seg — 38.1 (2019-04-30) WSMA-Seg — 38.1 (2019-04-30) ResNeXt-64x4d-101 NAS-FCOS @128-256 w/improvements — 46.1 (2019-06-11) ResNet-50 NAS-FCOS @256 — 39.8 (2019-06-11) ResNeXt-64x4d-101 NAS-FCOS @128-256 w/improvements — 46.1 (2019-06-11) ResNet-50 NAS-FCOS @256 — 39.8 (2019-06-11) ResNeXt-64x4d-101 NAS-FCOS @128-256 w/improvements — 46.1 (2019-06-11) ResNet-50 NAS-FCOS @256 — 39.8 (2019-06-11) ResNeXt-64x4d-101 NAS-FCOS @128-256 w/improvements — 46.1 (2019-06-11) ResNet-50 NAS-FCOS @256 — 39.8 (2019-06-11) ResNeXt-64x4d-101 NAS-FCOS @128-256 w/improvements — 46.1 (2019-06-11) ResNet-50 NAS-FCOS @256 — 39.8 (2019-06-11) Cascade R-CNN — 42.8 (2019-06-24) Cascade R-CNN — 42.8 (2019-06-24) Cascade R-CNN — 42.8 (2019-06-24) Cascade R-CNN — 42.8 (2019-06-24) Cascade R-CNN — 42.8 (2019-06-24) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) Faster R-CNN + FPN + CGD — 37.9 (2019-07-23) Faster R-CNN + FPN + CGD — 37.9 (2019-07-23) Faster R-CNN + FPN + CGD — 37.9 (2019-07-23) Faster R-CNN + FPN + CGD — 37.9 (2019-07-23) Faster R-CNN + FPN + CGD — 37.9 (2019-07-23) Faster R-CNN (LIP-ResNet-101-MD w FPN) — 43.9 (2019-08-12) Faster R-CNN (LIP-ResNet-101-MD w FPN) — 43.9 (2019-08-12) Faster R-CNN (LIP-ResNet-101-MD w FPN) — 43.9 (2019-08-12) Faster R-CNN (LIP-ResNet-101-MD w FPN) — 43.9 (2019-08-12) Faster R-CNN (LIP-ResNet-101-MD w FPN) — 43.9 (2019-08-12) MatrixNet Corners (ResNet-152, multi-scale) — 47.8 (2019-08-13) MatrixNet Corners (ResNet-152, multi-scale) — 47.8 (2019-08-13) MatrixNet Corners (ResNet-152, multi-scale) — 47.8 (2019-08-13) MatrixNet Corners (ResNet-152, multi-scale) — 47.8 (2019-08-13) MatrixNet Corners (ResNet-152, multi-scale) — 47.8 (2019-08-13) HTC (HRNetV2p-W48) — 47.3 (2019-08-20) Mask R-CNN (HRNetV2p-W48 + cascade) — 46.1 (2019-08-20) CenterNet (HRNetV2-W48) — 43.5 (2019-08-20) Faster R-CNN (HRNetV2p-W48) — 42.4 (2019-08-20) FCOS (HRNetV2p-W48) — 40.5 (2019-08-20) HTC (HRNetV2p-W48) — 47.3 (2019-08-20) Mask R-CNN (HRNetV2p-W48 + cascade) — 46.1 (2019-08-20) CenterNet (HRNetV2-W48) — 43.5 (2019-08-20) Faster R-CNN (HRNetV2p-W48) — 42.4 (2019-08-20) FCOS (HRNetV2p-W48) — 40.5 (2019-08-20) HTC (HRNetV2p-W48) — 47.3 (2019-08-20) Mask R-CNN (HRNetV2p-W48 + cascade) — 46.1 (2019-08-20) CenterNet (HRNetV2-W48) — 43.5 (2019-08-20) Faster R-CNN (HRNetV2p-W48) — 42.4 (2019-08-20) FCOS (HRNetV2p-W48) — 40.5 (2019-08-20) HTC (HRNetV2p-W48) — 47.3 (2019-08-20) Mask R-CNN (HRNetV2p-W48 + cascade) — 46.1 (2019-08-20) CenterNet (HRNetV2-W48) — 43.5 (2019-08-20) Faster R-CNN (HRNetV2p-W48) — 42.4 (2019-08-20) FCOS (HRNetV2p-W48) — 40.5 (2019-08-20) HTC (HRNetV2p-W48) — 47.3 (2019-08-20) Mask R-CNN (HRNetV2p-W48 + cascade) — 46.1 (2019-08-20) CenterNet (HRNetV2-W48) — 43.5 (2019-08-20) Faster R-CNN (HRNetV2p-W48) — 42.4 (2019-08-20) FCOS (HRNetV2p-W48) — 40.5 (2019-08-20) Cascade R-CNN-FPN (ResNet-101, map-guided) — 45.9 (2019-08-21) Cascade R-CNN-FPN (ResNet-101, map-guided) — 45.9 (2019-08-21) Cascade R-CNN-FPN (ResNet-101, map-guided) — 45.9 (2019-08-21) Cascade R-CNN-FPN (ResNet-101, map-guided) — 45.9 (2019-08-21) Cascade R-CNN-FPN (ResNet-101, map-guided) — 45.9 (2019-08-21) FreeAnchor (ResNeXt-101) — 44.8 (2019-09-05) FreeAnchor (ResNeXt-101) — 44.8 (2019-09-05) FreeAnchor (ResNeXt-101) — 44.8 (2019-09-05) FreeAnchor (ResNeXt-101) — 44.8 (2019-09-05) FreeAnchor (ResNeXt-101) — 44.8 (2019-09-05) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) Faster R-CNN (Cascade RPN) — 40.6 (2019-09-15) Fast R-CNN (Cascade RPN) — 40.1 (2019-09-15) Faster R-CNN (Cascade RPN) — 40.6 (2019-09-15) Fast R-CNN (Cascade RPN) — 40.1 (2019-09-15) Faster R-CNN (Cascade RPN) — 40.6 (2019-09-15) Fast R-CNN (Cascade RPN) — 40.1 (2019-09-15) Faster R-CNN (Cascade RPN) — 40.6 (2019-09-15) Fast R-CNN (Cascade RPN) — 40.1 (2019-09-15) Faster R-CNN (Cascade RPN) — 40.6 (2019-09-15) Fast R-CNN (Cascade RPN) — 40.1 (2019-09-15) HSD (Rest101, 768x768, single-scale test) — 42.3 (2019-10-01) HSD (Rest101, 768x768, single-scale test) — 42.3 (2019-10-01) HSD (Rest101, 768x768, single-scale test) — 42.3 (2019-10-01) HSD (Rest101, 768x768, single-scale test) — 42.3 (2019-10-01) HSD (Rest101, 768x768, single-scale test) — 42.3 (2019-10-01) ResNet-50-DW-DPN (Deformable Kernels) — 40.6 (2019-10-07) ResNet-50-DW-DPN (Deformable Kernels) — 40.6 (2019-10-07) ResNet-50-DW-DPN (Deformable Kernels) — 40.6 (2019-10-07) ResNet-50-DW-DPN (Deformable Kernels) — 40.6 (2019-10-07) ResNet-50-DW-DPN (Deformable Kernels) — 40.6 (2019-10-07) CenterMask+VoVNetV2-99 (single-scale) — 45.8 (2019-11-15) CenterMask+VoVNet2-57 (single-scale) — 44.7 (2019-11-15) CenterMask + X-101-32x8d (single-scale) — 44.6 (2019-11-15) CenterMask+VoVNetV2-99 (single-scale) — 45.8 (2019-11-15) CenterMask+VoVNet2-57 (single-scale) — 44.7 (2019-11-15) CenterMask + X-101-32x8d (single-scale) — 44.6 (2019-11-15) CenterMask+VoVNetV2-99 (single-scale) — 45.8 (2019-11-15) CenterMask+VoVNet2-57 (single-scale) — 44.7 (2019-11-15) CenterMask + X-101-32x8d (single-scale) — 44.6 (2019-11-15) CenterMask+VoVNetV2-99 (single-scale) — 45.8 (2019-11-15) CenterMask+VoVNet2-57 (single-scale) — 44.7 (2019-11-15) CenterMask + X-101-32x8d (single-scale) — 44.6 (2019-11-15) CenterMask+VoVNetV2-99 (single-scale) — 45.8 (2019-11-15) CenterMask+VoVNet2-57 (single-scale) — 44.7 (2019-11-15) CenterMask + X-101-32x8d (single-scale) — 44.6 (2019-11-15) EfficientDet-D7 (1536) — 52.6 (2019-11-20) EfficientDet-D7 (1536) — 52.6 (2019-11-20) EfficientDet-D7 (1536) — 52.6 (2019-11-20) EfficientDet-D7 (1536) — 52.6 (2019-11-20) EfficientDet-D7 (1536) — 52.6 (2019-11-20) YOLOv3 @800 + ASFF* (Darknet-53) — 43.9 (2019-11-21) YOLOv3 @800 + ASFF* (Darknet-53) — 43.9 (2019-11-21) YOLOv3 @800 + ASFF* (Darknet-53) — 43.9 (2019-11-21) YOLOv3 @800 + ASFF* (Darknet-53) — 43.9 (2019-11-21) YOLOv3 @800 + ASFF* (Darknet-53) — 43.9 (2019-11-21) SAPD (ResNeXt-101, single-scale) — 47.4 (2019-11-27) SAPD (ResNeXt-101, single-scale) — 47.4 (2019-11-27) SAPD (ResNeXt-101, single-scale) — 47.4 (2019-11-27) SAPD (ResNeXt-101, single-scale) — 47.4 (2019-11-27) SAPD (ResNeXt-101, single-scale) — 47.4 (2019-11-27) MAL (ResNeXt101, multi-scale) — 47.0 (2019-12-04) MAL (ResNeXt101, single-scale) — 45.9 (2019-12-04) MAL (ResNet50, single-scale) — 39.2 (2019-12-04) MAL (ResNeXt101, multi-scale) — 47.0 (2019-12-04) MAL (ResNeXt101, single-scale) — 45.9 (2019-12-04) MAL (ResNet50, single-scale) — 39.2 (2019-12-04) MAL (ResNeXt101, multi-scale) — 47.0 (2019-12-04) MAL (ResNeXt101, single-scale) — 45.9 (2019-12-04) MAL (ResNet50, single-scale) — 39.2 (2019-12-04) MAL (ResNeXt101, multi-scale) — 47.0 (2019-12-04) MAL (ResNeXt101, single-scale) — 45.9 (2019-12-04) MAL (ResNet50, single-scale) — 39.2 (2019-12-04) MAL (ResNeXt101, multi-scale) — 47.0 (2019-12-04) MAL (ResNeXt101, single-scale) — 45.9 (2019-12-04) MAL (ResNet50, single-scale) — 39.2 (2019-12-04) ATSS (ResNetXt-64x4d-101+DCN,multi-scale) — 50.7 (2019-12-05) ATSS (ResNetXt-64x4d-101+DCN,multi-scale) — 50.7 (2019-12-05) ATSS (ResNetXt-64x4d-101+DCN,multi-scale) — 50.7 (2019-12-05) ATSS (ResNetXt-64x4d-101+DCN,multi-scale) — 50.7 (2019-12-05) ATSS (ResNetXt-64x4d-101+DCN,multi-scale) — 50.7 (2019-12-05) RetinaNet (SpineNet-190, 1280x1280) — 52.1 (2019-12-10) RetinaNet (SpineNet-143, 1280x1280) — 50.7 (2019-12-10) RetinaNet (SpineNet-96, 1024x1024) — 48.6 (2019-12-10) RetinaNet (SpineNet-49, 896x896) — 46.7 (2019-12-10) RetinaNet (SpineNet-49, 640x640) — 44.3 (2019-12-10) SpineNet-49 (640, RetinaNet, single-scale) — 42.8 (2019-12-10) RetinaNet (SpineNet-49S, 640x640) — 41.5 (2019-12-10) RetinaNet (SpineNet-190, 1280x1280) — 52.1 (2019-12-10) RetinaNet (SpineNet-143, 1280x1280) — 50.7 (2019-12-10) RetinaNet (SpineNet-96, 1024x1024) — 48.6 (2019-12-10) RetinaNet (SpineNet-49, 896x896) — 46.7 (2019-12-10) RetinaNet (SpineNet-49, 640x640) — 44.3 (2019-12-10) SpineNet-49 (640, RetinaNet, single-scale) — 42.8 (2019-12-10) RetinaNet (SpineNet-49S, 640x640) — 41.5 (2019-12-10) RetinaNet (SpineNet-190, 1280x1280) — 52.1 (2019-12-10) RetinaNet (SpineNet-143, 1280x1280) — 50.7 (2019-12-10) RetinaNet (SpineNet-96, 1024x1024) — 48.6 (2019-12-10) RetinaNet (SpineNet-49, 896x896) — 46.7 (2019-12-10) RetinaNet (SpineNet-49, 640x640) — 44.3 (2019-12-10) SpineNet-49 (640, RetinaNet, single-scale) — 42.8 (2019-12-10) RetinaNet (SpineNet-49S, 640x640) — 41.5 (2019-12-10) RetinaNet (SpineNet-190, 1280x1280) — 52.1 (2019-12-10) RetinaNet (SpineNet-143, 1280x1280) — 50.7 (2019-12-10) RetinaNet (SpineNet-96, 1024x1024) — 48.6 (2019-12-10) RetinaNet (SpineNet-49, 896x896) — 46.7 (2019-12-10) RetinaNet (SpineNet-49, 640x640) — 44.3 (2019-12-10) SpineNet-49 (640, RetinaNet, single-scale) — 42.8 (2019-12-10) RetinaNet (SpineNet-49S, 640x640) — 41.5 (2019-12-10) RetinaNet (SpineNet-190, 1280x1280) — 52.1 (2019-12-10) RetinaNet (SpineNet-143, 1280x1280) — 50.7 (2019-12-10) RetinaNet (SpineNet-96, 1024x1024) — 48.6 (2019-12-10) RetinaNet (SpineNet-49, 896x896) — 46.7 (2019-12-10) RetinaNet (SpineNet-49, 640x640) — 44.3 (2019-12-10) SpineNet-49 (640, RetinaNet, single-scale) — 42.8 (2019-12-10) RetinaNet (SpineNet-49S, 640x640) — 41.5 (2019-12-10) RDSNet (ResNet-101, RetinaNet, mask, MBRM) — 40.3 (2019-12-11) RDSNet (ResNet-101, RetinaNet, mask, MBRM) — 40.3 (2019-12-11) RDSNet (ResNet-101, RetinaNet, mask, MBRM) — 40.3 (2019-12-11) RDSNet (ResNet-101, RetinaNet, mask, MBRM) — 40.3 (2019-12-11) RDSNet (ResNet-101, RetinaNet, mask, MBRM) — 40.3 (2019-12-11) Mask R-CNN (ResNet-101-FPN, CBN) — 40.1 (2020-02-13) Mask R-CNN (ResNet-101-FPN, CBN) — 40.1 (2020-02-13) Mask R-CNN (ResNet-101-FPN, CBN) — 40.1 (2020-02-13) Mask R-CNN (ResNet-101-FPN, CBN) — 40.1 (2020-02-13) Mask R-CNN (ResNet-101-FPN, CBN) — 40.1 (2020-02-13) TSD(SENet154-DCN,multi-scale) — 51.2 (2020-03-17) TSD(ResNet-101-Deformable, Image Pyramid) — 49.4 (2020-03-17) TSD(SENet154-DCN,multi-scale) — 51.2 (2020-03-17) TSD(ResNet-101-Deformable, Image Pyramid) — 49.4 (2020-03-17) TSD(SENet154-DCN,multi-scale) — 51.2 (2020-03-17) TSD(ResNet-101-Deformable, Image Pyramid) — 49.4 (2020-03-17) TSD(SENet154-DCN,multi-scale) — 51.2 (2020-03-17) TSD(ResNet-101-Deformable, Image Pyramid) — 49.4 (2020-03-17) TSD(SENet154-DCN,multi-scale) — 51.2 (2020-03-17) TSD(ResNet-101-Deformable, Image Pyramid) — 49.4 (2020-03-17) SaccadeNet (DLA-34-DCN) — 38.5 (2020-03-26) SaccadeNet (DLA-34-DCN) — 38.5 (2020-03-26) SaccadeNet (DLA-34-DCN) — 38.5 (2020-03-26) SaccadeNet (DLA-34-DCN) — 38.5 (2020-03-26) SaccadeNet (DLA-34-DCN) — 38.5 (2020-03-26) Dynamic R-CNN (ResNet-101-DCN, multi-scale) — 50.1 (2020-04-13) Dynamic R-CNN (ResNet-101-DCN, multi-scale) — 50.1 (2020-04-13) Dynamic R-CNN (ResNet-101-DCN, multi-scale) — 50.1 (2020-04-13) Dynamic R-CNN (ResNet-101-DCN, multi-scale) — 50.1 (2020-04-13) Dynamic R-CNN (ResNet-101-DCN, multi-scale) — 50.1 (2020-04-13) ResNeSt-200 (multi-scale) — 53.3 (2020-04-19) ResNeSt-200 (multi-scale) — 53.3 (2020-04-19) ResNeSt-200 (multi-scale) — 53.3 (2020-04-19) ResNeSt-200 (multi-scale) — 53.3 (2020-04-19) ResNeSt-200 (multi-scale) — 53.3 (2020-04-19) YOLOv4-608 — 43.5 (2020-04-23) YOLOv4-608 — 43.5 (2020-04-23) YOLOv4-608 — 43.5 (2020-04-23) YOLOv4-608 — 43.5 (2020-04-23) YOLOv4-608 — 43.5 (2020-04-23) FreeAnchor + SEPC (DCN, ResNext-101-64x4d) — 50.1 (2020-05-06) FreeAnchor + SEPC (DCN, ResNext-101-64x4d) — 50.1 (2020-05-06) FreeAnchor + SEPC (DCN, ResNext-101-64x4d) — 50.1 (2020-05-06) FreeAnchor + SEPC (DCN, ResNext-101-64x4d) — 50.1 (2020-05-06) FreeAnchor + SEPC (DCN, ResNext-101-64x4d) — 50.1 (2020-05-06) AC-FPN Cascade R-CNN (X-152-32x8d-FPN-IN5k, multi scale, only CEM) — 51.9 (2020-05-23) AC-FPN Cascade R-CNN(ResNet-101, single scale) — 45.0 (2020-05-23) AC-FPN Cascade R-CNN (X-152-32x8d-FPN-IN5k, multi scale, only CEM) — 51.9 (2020-05-23) AC-FPN Cascade R-CNN(ResNet-101, single scale) — 45.0 (2020-05-23) AC-FPN Cascade R-CNN (X-152-32x8d-FPN-IN5k, multi scale, only CEM) — 51.9 (2020-05-23) AC-FPN Cascade R-CNN(ResNet-101, single scale) — 45.0 (2020-05-23) AC-FPN Cascade R-CNN (X-152-32x8d-FPN-IN5k, multi scale, only CEM) — 51.9 (2020-05-23) AC-FPN Cascade R-CNN(ResNet-101, single scale) — 45.0 (2020-05-23) AC-FPN Cascade R-CNN (X-152-32x8d-FPN-IN5k, multi scale, only CEM) — 51.9 (2020-05-23) AC-FPN Cascade R-CNN(ResNet-101, single scale) — 45.0 (2020-05-23) D2Det (ResNet-101-DCN, multi-scale test) — 50.1 (2020-06-01) D2Det (ResNet-101-DCN, multi-scale test) — 50.1 (2020-06-01) D2Det (ResNet-101-DCN, multi-scale test) — 50.1 (2020-06-01) D2Det (ResNet-101-DCN, multi-scale test) — 50.1 (2020-06-01) D2Det (ResNet-101-DCN, multi-scale test) — 50.1 (2020-06-01) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, multi-scale) — 54.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, single-scale) — 53.3 (2020-06-03) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, multi-scale) — 54.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, single-scale) — 53.3 (2020-06-03) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, multi-scale) — 54.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, single-scale) — 53.3 (2020-06-03) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, multi-scale) — 54.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, single-scale) — 53.3 (2020-06-03) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, multi-scale) — 54.7 (2020-06-03) DetectoRS (ResNeXt-101-32x4d, single-scale) — 53.3 (2020-06-03) GFL (X-101-32x4d-DCN, single-scale) — 48.2 (2020-06-08) GFL (X-101-32x4d-DCN, single-scale) — 48.2 (2020-06-08) GFL (X-101-32x4d-DCN, single-scale) — 48.2 (2020-06-08) GFL (X-101-32x4d-DCN, single-scale) — 48.2 (2020-06-08) GFL (X-101-32x4d-DCN, single-scale) — 48.2 (2020-06-08) SpineNet-190 (1280, with Self-training on OpenImages, single-scale) — 54.3 (2020-06-11) SpineNet-190 (1280, with Self-training on OpenImages, single-scale) — 54.3 (2020-06-11) SpineNet-190 (1280, with Self-training on OpenImages, single-scale) — 54.3 (2020-06-11) SpineNet-190 (1280, with Self-training on OpenImages, single-scale) — 54.3 (2020-06-11) SpineNet-190 (1280, with Self-training on OpenImages, single-scale) — 54.3 (2020-06-11) Gaussian-FCOS — 46.0 (2020-06-28) Gaussian-FCOS — 46.0 (2020-06-28) Gaussian-FCOS — 46.0 (2020-06-28) Gaussian-FCOS — 46.0 (2020-06-28) Gaussian-FCOS — 46.0 (2020-06-28) HoughNet (MS) — 46.4 (2020-07-05) HoughNet (MS) — 46.4 (2020-07-05) HoughNet (MS) — 46.4 (2020-07-05) HoughNet (MS) — 46.4 (2020-07-05) HoughNet (MS) — 46.4 (2020-07-05) PAA (ResNext-152-32x8d + DCN, multi-scale) — 53.5 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN, multi-scale) — 52.1 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN) — 49.4 (2020-07-16) PAA (ResNext-152-32x8d + DCN, multi-scale) — 53.5 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN, multi-scale) — 52.1 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN) — 49.4 (2020-07-16) PAA (ResNext-152-32x8d + DCN, multi-scale) — 53.5 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN, multi-scale) — 52.1 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN) — 49.4 (2020-07-16) PAA (ResNext-152-32x8d + DCN, multi-scale) — 53.5 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN, multi-scale) — 52.1 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN) — 49.4 (2020-07-16) PAA (ResNext-152-32x8d + DCN, multi-scale) — 53.5 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN, multi-scale) — 52.1 (2020-07-16) RepPoints v2 (ResNeXt-101, DCN) — 49.4 (2020-07-16) CPNDet (Hourglass-104, multi-scale) — 49.2 (2020-07-27) CPNDet (Hourglass-104, multi-scale) — 49.2 (2020-07-27) CPNDet (Hourglass-104, multi-scale) — 49.2 (2020-07-27) CPNDet (Hourglass-104, multi-scale) — 49.2 (2020-07-27) CPNDet (Hourglass-104, multi-scale) — 49.2 (2020-07-27) PPDet (ResNeXt-101-FPN, multiscale) — 46.3 (2020-08-03) PPDet (ResNeXt-101-FPN, multiscale) — 46.3 (2020-08-03) PPDet (ResNeXt-101-FPN, multiscale) — 46.3 (2020-08-03) PPDet (ResNeXt-101-FPN, multiscale) — 46.3 (2020-08-03) PPDet (ResNeXt-101-FPN, multiscale) — 46.3 (2020-08-03) aLRP Loss (ResNext-101-64x4d, DCN, multiscale test) — 50.2 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, single scale) — 48.9 (2020-09-28) aLRP Loss (ResNext-101-64x4d, single scale) — 47.8 (2020-09-28) aLRP Loss (ResNext-101, DCN, 500 scale) — 44.6 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, multiscale test) — 50.2 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, single scale) — 48.9 (2020-09-28) aLRP Loss (ResNext-101-64x4d, single scale) — 47.8 (2020-09-28) aLRP Loss (ResNext-101, DCN, 500 scale) — 44.6 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, multiscale test) — 50.2 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, single scale) — 48.9 (2020-09-28) aLRP Loss (ResNext-101-64x4d, single scale) — 47.8 (2020-09-28) aLRP Loss (ResNext-101, DCN, 500 scale) — 44.6 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, multiscale test) — 50.2 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, single scale) — 48.9 (2020-09-28) aLRP Loss (ResNext-101-64x4d, single scale) — 47.8 (2020-09-28) aLRP Loss (ResNext-101, DCN, 500 scale) — 44.6 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, multiscale test) — 50.2 (2020-09-28) aLRP Loss (ResNext-101-64x4d, DCN, single scale) — 48.9 (2020-09-28) aLRP Loss (ResNext-101-64x4d, single scale) — 47.8 (2020-09-28) aLRP Loss (ResNext-101, DCN, 500 scale) — 44.6 (2020-09-28) Deformable DETR (ResNeXt-101+DCN) — 52.3 (2020-10-08) Deformable DETR (ResNeXt-101+DCN) — 52.3 (2020-10-08) Deformable DETR (ResNeXt-101+DCN) — 52.3 (2020-10-08) Deformable DETR (ResNeXt-101+DCN) — 52.3 (2020-10-08) Deformable DETR (ResNeXt-101+DCN) — 52.3 (2020-10-08) RelationNet++ (ResNeXt-64x4d-101-DCN) — 52.7 (2020-10-29) RelationNet++ (ResNeXt-64x4d-101-DCN) — 52.7 (2020-10-29) RelationNet++ (ResNeXt-64x4d-101-DCN) — 52.7 (2020-10-29) RelationNet++ (ResNeXt-64x4d-101-DCN) — 52.7 (2020-10-29) RelationNet++ (ResNeXt-64x4d-101-DCN) — 52.7 (2020-10-29) YOLOv4-P7 with TTA — 55.8 (2020-11-16) YOLOv4-P6 with TTA — 54.9 (2020-11-16) YOLOv4-P6 CSP-P6 (single-scale, 32 fps) — 54.3 (2020-11-16) YOLOv4-P5 with TTA — 52.5 (2020-11-16) YOLOv4 (CD53) — 45.5 (2020-11-16) YOLOv4-P7 with TTA — 55.8 (2020-11-16) YOLOv4-P6 with TTA — 54.9 (2020-11-16) YOLOv4-P6 CSP-P6 (single-scale, 32 fps) — 54.3 (2020-11-16) YOLOv4-P5 with TTA — 52.5 (2020-11-16) YOLOv4 (CD53) — 45.5 (2020-11-16) YOLOv4-P7 with TTA — 55.8 (2020-11-16) YOLOv4-P6 with TTA — 54.9 (2020-11-16) YOLOv4-P6 CSP-P6 (single-scale, 32 fps) — 54.3 (2020-11-16) YOLOv4-P5 with TTA — 52.5 (2020-11-16) YOLOv4 (CD53) — 45.5 (2020-11-16) YOLOv4-P7 with TTA — 55.8 (2020-11-16) YOLOv4-P6 with TTA — 54.9 (2020-11-16) YOLOv4-P6 CSP-P6 (single-scale, 32 fps) — 54.3 (2020-11-16) YOLOv4-P5 with TTA — 52.5 (2020-11-16) YOLOv4 (CD53) — 45.5 (2020-11-16) YOLOv4-P7 with TTA — 55.8 (2020-11-16) YOLOv4-P6 with TTA — 54.9 (2020-11-16) YOLOv4-P6 CSP-P6 (single-scale, 32 fps) — 54.3 (2020-11-16) YOLOv4-P5 with TTA — 52.5 (2020-11-16) YOLOv4 (CD53) — 45.5 (2020-11-16) GFLV2 (Res2Net-101, DCN, multiscale) — 53.3 (2020-11-25) GFLV2 (Res2Net-101, DCN) — 50.6 (2020-11-25) GFLV2 (ResNeXt-101, 32x4d, DCN) — 49.0 (2020-11-25) GFLV2 (ResNet-101-DCN) — 48.3 (2020-11-25) GFLV2 (ResNet-101) — 46.2 (2020-11-25) GFLV2 (ResNet-50) — 44.3 (2020-11-25) Mask R-CNN (Bottleneck-injected ResNet-50, FPN) — 36.9 (2020-11-25) Faster R-CNN (Bottleneck-injected ResNet-50 and FPN) — 35.9 (2020-11-25) GFLV2 (Res2Net-101, DCN, multiscale) — 53.3 (2020-11-25) GFLV2 (Res2Net-101, DCN) — 50.6 (2020-11-25) GFLV2 (ResNeXt-101, 32x4d, DCN) — 49.0 (2020-11-25) GFLV2 (ResNet-101-DCN) — 48.3 (2020-11-25) GFLV2 (ResNet-101) — 46.2 (2020-11-25) GFLV2 (ResNet-50) — 44.3 (2020-11-25) Mask R-CNN (Bottleneck-injected ResNet-50, FPN) — 36.9 (2020-11-25) Faster R-CNN (Bottleneck-injected ResNet-50 and FPN) — 35.9 (2020-11-25) GFLV2 (Res2Net-101, DCN, multiscale) — 53.3 (2020-11-25) GFLV2 (Res2Net-101, DCN) — 50.6 (2020-11-25) GFLV2 (ResNeXt-101, 32x4d, DCN) — 49.0 (2020-11-25) GFLV2 (ResNet-101-DCN) — 48.3 (2020-11-25) GFLV2 (ResNet-101) — 46.2 (2020-11-25) GFLV2 (ResNet-50) — 44.3 (2020-11-25) Mask R-CNN (Bottleneck-injected ResNet-50, FPN) — 36.9 (2020-11-25) Faster R-CNN (Bottleneck-injected ResNet-50 and FPN) — 35.9 (2020-11-25) GFLV2 (Res2Net-101, DCN, multiscale) — 53.3 (2020-11-25) GFLV2 (Res2Net-101, DCN) — 50.6 (2020-11-25) GFLV2 (ResNeXt-101, 32x4d, DCN) — 49.0 (2020-11-25) GFLV2 (ResNet-101-DCN) — 48.3 (2020-11-25) GFLV2 (ResNet-101) — 46.2 (2020-11-25) GFLV2 (ResNet-50) — 44.3 (2020-11-25) Mask R-CNN (Bottleneck-injected ResNet-50, FPN) — 36.9 (2020-11-25) Faster R-CNN (Bottleneck-injected ResNet-50 and FPN) — 35.9 (2020-11-25) GFLV2 (Res2Net-101, DCN, multiscale) — 53.3 (2020-11-25) GFLV2 (Res2Net-101, DCN) — 50.6 (2020-11-25) GFLV2 (ResNeXt-101, 32x4d, DCN) — 49.0 (2020-11-25) GFLV2 (ResNet-101-DCN) — 48.3 (2020-11-25) GFLV2 (ResNet-101) — 46.2 (2020-11-25) GFLV2 (ResNet-50) — 44.3 (2020-11-25) Mask R-CNN (Bottleneck-injected ResNet-50, FPN) — 36.9 (2020-11-25) Faster R-CNN (Bottleneck-injected ResNet-50 and FPN) — 35.9 (2020-11-25) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280) — 54.8 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280) — 54.8 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280) — 54.8 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280) — 54.8 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Cascade Eff-B7 NAS-FPN (1280) — 54.8 (2020-12-13) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 52.3 (2020-12-24) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 52.3 (2020-12-24) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 52.3 (2020-12-24) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 52.3 (2020-12-24) GCNet (ResNeXt-101 + DCN + cascade + GC r4) — 52.3 (2020-12-24) CenterNet2 (Res2Net-101-DCN-BiFPN, self-training, 1560 single-scale) — 56.4 (2021-03-12) CenterNet2 (Res2Net-101-DCN-BiFPN, self-training, 1560 single-scale) — 56.4 (2021-03-12) CenterNet2 (Res2Net-101-DCN-BiFPN, self-training, 1560 single-scale) — 56.4 (2021-03-12) CenterNet2 (Res2Net-101-DCN-BiFPN, self-training, 1560 single-scale) — 56.4 (2021-03-12) CenterNet2 (Res2Net-101-DCN-BiFPN, self-training, 1560 single-scale) — 56.4 (2021-03-12) YOLOF-DC5 — 44.3 (2021-03-17) YOLOF-DC5 — 44.3 (2021-03-17) YOLOF-DC5 — 44.3 (2021-03-17) YOLOF-DC5 — 44.3 (2021-03-17) YOLOF-DC5 — 44.3 (2021-03-17) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) Swin-L (HTC++, single scale) — 57.7 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, multi-scale) — 54.1 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, single-scale) — 51.3 (2021-03-25) UniverseNet-20.08 (Res2Net-50, DCN, single-scale) — 48.8 (2021-03-25) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) Swin-L (HTC++, single scale) — 57.7 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, multi-scale) — 54.1 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, single-scale) — 51.3 (2021-03-25) UniverseNet-20.08 (Res2Net-50, DCN, single-scale) — 48.8 (2021-03-25) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) Swin-L (HTC++, single scale) — 57.7 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, multi-scale) — 54.1 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, single-scale) — 51.3 (2021-03-25) UniverseNet-20.08 (Res2Net-50, DCN, single-scale) — 48.8 (2021-03-25) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) Swin-L (HTC++, single scale) — 57.7 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, multi-scale) — 54.1 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, single-scale) — 51.3 (2021-03-25) UniverseNet-20.08 (Res2Net-50, DCN, single-scale) — 48.8 (2021-03-25) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) Swin-L (HTC++, single scale) — 57.7 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, multi-scale) — 54.1 (2021-03-25) UniverseNet-20.08d (Res2Net-101, DCN, single-scale) — 51.3 (2021-03-25) UniverseNet-20.08 (Res2Net-50, DCN, single-scale) — 48.8 (2021-03-25) OTA (ResNeXt-101+DCN, multiscale) — 51.5 (2021-03-26) OTA (ResNeXt-101+DCN, multiscale) — 51.5 (2021-03-26) OTA (ResNeXt-101+DCN, multiscale) — 51.5 (2021-03-26) OTA (ResNeXt-101+DCN, multiscale) — 51.5 (2021-03-26) OTA (ResNeXt-101+DCN, multiscale) — 51.5 (2021-03-26) LSNet (Res2Net-101+ DCN, multi-scale) — 53.5 (2021-04-11) LSNet (Res2Net-101+ DCN, multi-scale) — 53.5 (2021-04-11) LSNet (Res2Net-101+ DCN, multi-scale) — 53.5 (2021-04-11) LSNet (Res2Net-101+ DCN, multi-scale) — 53.5 (2021-04-11) LSNet (Res2Net-101+ DCN, multi-scale) — 53.5 (2021-04-11) ISTR (ResNet50-FPN-3x, single-scale) — 56.4 (2021-05-03) ISTR (ResNet101-FPN-3x, single-scale) — 48.1 (2021-05-03) ISTR (ResNet50-FPN-3x) — 46.8 (2021-05-03) ISTR (ResNet50-FPN-3x, single-scale) — 56.4 (2021-05-03) ISTR (ResNet101-FPN-3x, single-scale) — 48.1 (2021-05-03) ISTR (ResNet50-FPN-3x) — 46.8 (2021-05-03) ISTR (ResNet50-FPN-3x, single-scale) — 56.4 (2021-05-03) ISTR (ResNet101-FPN-3x, single-scale) — 48.1 (2021-05-03) ISTR (ResNet50-FPN-3x) — 46.8 (2021-05-03) ISTR (ResNet50-FPN-3x, single-scale) — 56.4 (2021-05-03) ISTR (ResNet101-FPN-3x, single-scale) — 48.1 (2021-05-03) ISTR (ResNet50-FPN-3x) — 46.8 (2021-05-03) ISTR (ResNet50-FPN-3x, single-scale) — 56.4 (2021-05-03) ISTR (ResNet101-FPN-3x, single-scale) — 48.1 (2021-05-03) ISTR (ResNet50-FPN-3x) — 46.8 (2021-05-03) QueryInst (single-scale) — 56.1 (2021-05-05) QueryInst (single-scale) — 56.1 (2021-05-05) QueryInst (single-scale) — 56.1 (2021-05-05) QueryInst (single-scale) — 56.1 (2021-05-05) QueryInst (single-scale) — 56.1 (2021-05-05) YOLOR-D6 (1280, single-scale, 30 fps) — 55.4 (2021-05-10) YOLOR-D6 (1280, single-scale, 30 fps) — 55.4 (2021-05-10) YOLOR-D6 (1280, single-scale, 30 fps) — 55.4 (2021-05-10) YOLOR-D6 (1280, single-scale, 30 fps) — 55.4 (2021-05-10) YOLOR-D6 (1280, single-scale, 30 fps) — 55.4 (2021-05-10) SOLQ (Swin-L, single scale) — 56.5 (2021-06-04) SOLQ (ResNet101, single scale) — 48.7 (2021-06-04) SOLQ (ResNet50, single scale) — 47.8 (2021-06-04) SOLQ (Swin-L, single scale) — 56.5 (2021-06-04) SOLQ (ResNet101, single scale) — 48.7 (2021-06-04) SOLQ (ResNet50, single scale) — 47.8 (2021-06-04) SOLQ (Swin-L, single scale) — 56.5 (2021-06-04) SOLQ (ResNet101, single scale) — 48.7 (2021-06-04) SOLQ (ResNet50, single scale) — 47.8 (2021-06-04) SOLQ (Swin-L, single scale) — 56.5 (2021-06-04) SOLQ (ResNet101, single scale) — 48.7 (2021-06-04) SOLQ (ResNet50, single scale) — 47.8 (2021-06-04) SOLQ (Swin-L, single scale) — 56.5 (2021-06-04) SOLQ (ResNet101, single scale) — 48.7 (2021-06-04) SOLQ (ResNet50, single scale) — 47.8 (2021-06-04) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) DyHead (Swin-L, multi scale) — 58.7 (2021-06-15) DyHead (ResNeXt-64x4d-101-DCN, multi scale) — 54.0 (2021-06-15) DyHead (ResNeXt-64x4d-101) — 47.7 (2021-06-15) DyHead (ResNet-50) — 43.0 (2021-06-15) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) DyHead (Swin-L, multi scale) — 58.7 (2021-06-15) DyHead (ResNeXt-64x4d-101-DCN, multi scale) — 54.0 (2021-06-15) DyHead (ResNeXt-64x4d-101) — 47.7 (2021-06-15) DyHead (ResNet-50) — 43.0 (2021-06-15) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) DyHead (Swin-L, multi scale) — 58.7 (2021-06-15) DyHead (ResNeXt-64x4d-101-DCN, multi scale) — 54.0 (2021-06-15) DyHead (ResNeXt-64x4d-101) — 47.7 (2021-06-15) DyHead (ResNet-50) — 43.0 (2021-06-15) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) DyHead (Swin-L, multi scale) — 58.7 (2021-06-15) DyHead (ResNeXt-64x4d-101-DCN, multi scale) — 54.0 (2021-06-15) DyHead (ResNeXt-64x4d-101) — 47.7 (2021-06-15) DyHead (ResNet-50) — 43.0 (2021-06-15) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) DyHead (Swin-L, multi scale) — 58.7 (2021-06-15) DyHead (ResNeXt-64x4d-101-DCN, multi scale) — 54.0 (2021-06-15) DyHead (ResNeXt-64x4d-101) — 47.7 (2021-06-15) DyHead (ResNet-50) — 43.0 (2021-06-15) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) CBNetV2 (Dual-Swin-L HTC, multi-scale) — 60.1 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, single-scale) — 59.4 (2021-07-01) Focal-L (DyHead, multi-scale) — 58.9 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, multi-scale) — 60.1 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, single-scale) — 59.4 (2021-07-01) Focal-L (DyHead, multi-scale) — 58.9 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, multi-scale) — 60.1 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, single-scale) — 59.4 (2021-07-01) Focal-L (DyHead, multi-scale) — 58.9 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, multi-scale) — 60.1 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, single-scale) — 59.4 (2021-07-01) Focal-L (DyHead, multi-scale) — 58.9 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, multi-scale) — 60.1 (2021-07-01) CBNetV2 (Dual-Swin-L HTC, single-scale) — 59.4 (2021-07-01) Focal-L (DyHead, multi-scale) — 58.9 (2021-07-01) YOLOX-x(Modified CSP v5, 640x640, single-scale) — 51.5 (2021-07-18) YOLOX-X (Modified CSP v5) — 51.2 (2021-07-18) YOLOX-Darknet53(Darknet53, 640x640, single-scale) — 48.0 (2021-07-18) YOLOX-x(Modified CSP v5, 640x640, single-scale) — 51.5 (2021-07-18) YOLOX-X (Modified CSP v5) — 51.2 (2021-07-18) YOLOX-Darknet53(Darknet53, 640x640, single-scale) — 48.0 (2021-07-18) YOLOX-x(Modified CSP v5, 640x640, single-scale) — 51.5 (2021-07-18) YOLOX-X (Modified CSP v5) — 51.2 (2021-07-18) YOLOX-Darknet53(Darknet53, 640x640, single-scale) — 48.0 (2021-07-18) YOLOX-x(Modified CSP v5, 640x640, single-scale) — 51.5 (2021-07-18) YOLOX-X (Modified CSP v5) — 51.2 (2021-07-18) YOLOX-Darknet53(Darknet53, 640x640, single-scale) — 48.0 (2021-07-18) YOLOX-x(Modified CSP v5, 640x640, single-scale) — 51.5 (2021-07-18) YOLOX-X (Modified CSP v5) — 51.2 (2021-07-18) YOLOX-Darknet53(Darknet53, 640x640, single-scale) — 48.0 (2021-07-18) TAL + TAP — 42.5 (2021-08-17) TAL + TAP — 42.5 (2021-08-17) TAL + TAP — 42.5 (2021-08-17) TAL + TAP — 42.5 (2021-08-17) TAL + TAP — 42.5 (2021-08-17) iBOT (ViT-B/16) — 51.2 (2021-11-15) iBOT (ViT-S/16) — 49.4 (2021-11-15) iBOT (ViT-B/16) — 51.2 (2021-11-15) iBOT (ViT-S/16) — 49.4 (2021-11-15) iBOT (ViT-B/16) — 51.2 (2021-11-15) iBOT (ViT-S/16) — 49.4 (2021-11-15) iBOT (ViT-B/16) — 51.2 (2021-11-15) iBOT (ViT-S/16) — 49.4 (2021-11-15) iBOT (ViT-B/16) — 51.2 (2021-11-15) iBOT (ViT-S/16) — 49.4 (2021-11-15) SwinV2-G (HTC++) — 63.1 (2021-11-18) SwinV2-G (HTC++) — 63.1 (2021-11-18) SwinV2-G (HTC++) — 63.1 (2021-11-18) SwinV2-G (HTC++) — 63.1 (2021-11-18) SwinV2-G (HTC++) — 63.1 (2021-11-18) Florence-CoSwin-H — 62.4 (2021-11-22) Florence-CoSwin-H — 62.4 (2021-11-22) Florence-CoSwin-H — 62.4 (2021-11-22) Florence-CoSwin-H — 62.4 (2021-11-22) Florence-CoSwin-H — 62.4 (2021-11-22) GLIP (Swin-L, multi-scale) — 61.5 (2021-12-07) GLIP (Swin-L, multi-scale) — 61.5 (2021-12-07) GLIP (Swin-L, multi-scale) — 61.5 (2021-12-07) GLIP (Swin-L, multi-scale) — 61.5 (2021-12-07) GLIP (Swin-L, multi-scale) — 61.5 (2021-12-07) DAT-S (RetinaNet) — 47.9 (2022-01-03) DAT-S (RetinaNet) — 47.9 (2022-01-03) DAT-S (RetinaNet) — 47.9 (2022-01-03) DAT-S (RetinaNet) — 47.9 (2022-01-03) DAT-S (RetinaNet) — 47.9 (2022-01-03) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) FocalNet-H (DINO) — 64.4 (2022-03-22) FocalNet-H (DINO) — 64.4 (2022-03-22) FocalNet-H (DINO) — 64.4 (2022-03-22) FocalNet-H (DINO) — 64.4 (2022-03-22) FocalNet-H (DINO) — 64.4 (2022-03-22) PP-YOLOE-x(CSPRepResNet-x, 640x640, single-scale ) — 52.2 (2022-03-30) PP-YOLOE-l(CSPRepResNet-l, 640x640, single-scale ) — 51.4 (2022-03-30) PP-YOLOE-m(CSPRepResNet-m, 640x640, single-scale ) — 48.9 (2022-03-30) PP-YOLOE-s(CSPRepResNet-s, 640x640, single-scale ) — 43.1 (2022-03-30) PP-YOLOE-x(CSPRepResNet-x, 640x640, single-scale ) — 52.2 (2022-03-30) PP-YOLOE-l(CSPRepResNet-l, 640x640, single-scale ) — 51.4 (2022-03-30) PP-YOLOE-m(CSPRepResNet-m, 640x640, single-scale ) — 48.9 (2022-03-30) PP-YOLOE-s(CSPRepResNet-s, 640x640, single-scale ) — 43.1 (2022-03-30) PP-YOLOE-x(CSPRepResNet-x, 640x640, single-scale ) — 52.2 (2022-03-30) PP-YOLOE-l(CSPRepResNet-l, 640x640, single-scale ) — 51.4 (2022-03-30) PP-YOLOE-m(CSPRepResNet-m, 640x640, single-scale ) — 48.9 (2022-03-30) PP-YOLOE-s(CSPRepResNet-s, 640x640, single-scale ) — 43.1 (2022-03-30) PP-YOLOE-x(CSPRepResNet-x, 640x640, single-scale ) — 52.2 (2022-03-30) PP-YOLOE-l(CSPRepResNet-l, 640x640, single-scale ) — 51.4 (2022-03-30) PP-YOLOE-m(CSPRepResNet-m, 640x640, single-scale ) — 48.9 (2022-03-30) PP-YOLOE-s(CSPRepResNet-s, 640x640, single-scale ) — 43.1 (2022-03-30) PP-YOLOE-x(CSPRepResNet-x, 640x640, single-scale ) — 52.2 (2022-03-30) PP-YOLOE-l(CSPRepResNet-l, 640x640, single-scale ) — 51.4 (2022-03-30) PP-YOLOE-m(CSPRepResNet-m, 640x640, single-scale ) — 48.9 (2022-03-30) PP-YOLOE-s(CSPRepResNet-s, 640x640, single-scale ) — 43.1 (2022-03-30) PyCenterNet (Swin-L, multi-scale) — 57.1 (2022-04-18) PyCenterNet (Swin-L, multi-scale) — 57.1 (2022-04-18) PyCenterNet (Swin-L, multi-scale) — 57.1 (2022-04-18) PyCenterNet (Swin-L, multi-scale) — 57.1 (2022-04-18) PyCenterNet (Swin-L, multi-scale) — 57.1 (2022-04-18) ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) — 60.9 (2022-05-17) ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) — 60.4 (2022-05-17) ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) — 60.9 (2022-05-17) ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) — 60.4 (2022-05-17) ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) — 60.9 (2022-05-17) ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) — 60.4 (2022-05-17) ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) — 60.9 (2022-05-17) ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) — 60.4 (2022-05-17) ViT-Adapter-L (HTC++, BEiTv2 pretrain, multi-scale) — 60.9 (2022-05-17) ViT-Adapter-L (HTC++, BEiT pretrain, multi-scale) — 60.4 (2022-05-17) FD-SwinV2-G — 64.2 (2022-05-27) A2MIM (ViT-B) — 49.4 (2022-05-27) A2MIM (ResNet-50 2x) — 39.8 (2022-05-27) FD-SwinV2-G — 64.2 (2022-05-27) A2MIM (ViT-B) — 49.4 (2022-05-27) A2MIM (ResNet-50 2x) — 39.8 (2022-05-27) FD-SwinV2-G — 64.2 (2022-05-27) A2MIM (ViT-B) — 49.4 (2022-05-27) A2MIM (ResNet-50 2x) — 39.8 (2022-05-27) FD-SwinV2-G — 64.2 (2022-05-27) A2MIM (ViT-B) — 49.4 (2022-05-27) A2MIM (ResNet-50 2x) — 39.8 (2022-05-27) FD-SwinV2-G — 64.2 (2022-05-27) A2MIM (ViT-B) — 49.4 (2022-05-27) A2MIM (ResNet-50 2x) — 39.8 (2022-05-27) GLIPv2 (CoSwin-H, multi-scale) — 62.4 (2022-06-12) GLIPv2 (CoSwin-H, multi-scale) — 62.4 (2022-06-12) GLIPv2 (CoSwin-H, multi-scale) — 62.4 (2022-06-12) GLIPv2 (CoSwin-H, multi-scale) — 62.4 (2022-06-12) GLIPv2 (CoSwin-H, multi-scale) — 62.4 (2022-06-12) Boosting R-CNN* — 50.7 (2022-06-28) Boosting R-CNN* — 50.7 (2022-06-28) Boosting R-CNN* — 50.7 (2022-06-28) Boosting R-CNN* — 50.7 (2022-06-28) Boosting R-CNN* — 50.7 (2022-06-28) YOLOv7-D6 (44 fps) — 56.6 (2022-07-06) YOLOv7-E6 (56 fps) — 56.0 (2022-07-06) YOLOv7-W6 (84 fps) — 54.9 (2022-07-06) YOLOv7-X (114 fps) — 53.1 (2022-07-06) YOLOv7 (161 fps) — 51.4 (2022-07-06) YOLOv7-D6 (44 fps) — 56.6 (2022-07-06) YOLOv7-E6 (56 fps) — 56.0 (2022-07-06) YOLOv7-W6 (84 fps) — 54.9 (2022-07-06) YOLOv7-X (114 fps) — 53.1 (2022-07-06) YOLOv7 (161 fps) — 51.4 (2022-07-06) YOLOv7-D6 (44 fps) — 56.6 (2022-07-06) YOLOv7-E6 (56 fps) — 56.0 (2022-07-06) YOLOv7-W6 (84 fps) — 54.9 (2022-07-06) YOLOv7-X (114 fps) — 53.1 (2022-07-06) YOLOv7 (161 fps) — 51.4 (2022-07-06) YOLOv7-D6 (44 fps) — 56.6 (2022-07-06) YOLOv7-E6 (56 fps) — 56.0 (2022-07-06) YOLOv7-W6 (84 fps) — 54.9 (2022-07-06) YOLOv7-X (114 fps) — 53.1 (2022-07-06) YOLOv7 (161 fps) — 51.4 (2022-07-06) YOLOv7-D6 (44 fps) — 56.6 (2022-07-06) YOLOv7-E6 (56 fps) — 56.0 (2022-07-06) YOLOv7-W6 (84 fps) — 54.9 (2022-07-06) YOLOv7-X (114 fps) — 53.1 (2022-07-06) YOLOv7 (161 fps) — 51.4 (2022-07-06) BEiT-3 — 63.7 (2022-08-22) BEiT-3 — 63.7 (2022-08-22) BEiT-3 — 63.7 (2022-08-22) BEiT-3 — 63.7 (2022-08-22) BEiT-3 — 63.7 (2022-08-22) dBOT ViT-L (CLIP) — 56.8 (2022-09-08) dBOT ViT-L — 56.1 (2022-09-08) dBOT ViT-B (CLIP) — 53.6 (2022-09-08) dBOT ViT-B — 53.5 (2022-09-08) dBOT ViT-L (CLIP) — 56.8 (2022-09-08) dBOT ViT-L — 56.1 (2022-09-08) dBOT ViT-B (CLIP) — 53.6 (2022-09-08) dBOT ViT-B — 53.5 (2022-09-08) dBOT ViT-L (CLIP) — 56.8 (2022-09-08) dBOT ViT-L — 56.1 (2022-09-08) dBOT ViT-B (CLIP) — 53.6 (2022-09-08) dBOT ViT-B — 53.5 (2022-09-08) dBOT ViT-L (CLIP) — 56.8 (2022-09-08) dBOT ViT-L — 56.1 (2022-09-08) dBOT ViT-B (CLIP) — 53.6 (2022-09-08) dBOT ViT-B — 53.5 (2022-09-08) dBOT ViT-L (CLIP) — 56.8 (2022-09-08) dBOT ViT-L — 56.1 (2022-09-08) dBOT ViT-B (CLIP) — 53.6 (2022-09-08) dBOT ViT-B — 53.5 (2022-09-08) Group DETR v2 — 64.5 (2022-11-07) Group DETR v2 — 64.5 (2022-11-07) Group DETR v2 — 64.5 (2022-11-07) Group DETR v2 — 64.5 (2022-11-07) Group DETR v2 — 64.5 (2022-11-07) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) InternImage-XL — 64.3 (2022-11-10) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) InternImage-XL — 64.3 (2022-11-10) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) InternImage-XL — 64.3 (2022-11-10) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) InternImage-XL — 64.3 (2022-11-10) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) InternImage-XL — 64.3 (2022-11-10) EVA — 64.7 (2022-11-14) EVA — 64.7 (2022-11-14) EVA — 64.7 (2022-11-14) EVA — 64.7 (2022-11-14) EVA — 64.7 (2022-11-14) M3I Pre-training (InternImage-H) — 65.4 (2022-11-17) M3I Pre-training (InternImage-H) — 65.4 (2022-11-17) M3I Pre-training (InternImage-H) — 65.4 (2022-11-17) M3I Pre-training (InternImage-H) — 65.4 (2022-11-17) M3I Pre-training (InternImage-H) — 65.4 (2022-11-17) Co-DETR — 66.0 (2022-11-22) Co-DETR (Swin-L) — 64.8 (2022-11-22) Co-DETR — 66.0 (2022-11-22) Co-DETR (Swin-L) — 64.8 (2022-11-22) Co-DETR — 66.0 (2022-11-22) Co-DETR (Swin-L) — 64.8 (2022-11-22) Co-DETR — 66.0 (2022-11-22) Co-DETR (Swin-L) — 64.8 (2022-11-22) Co-DETR — 66.0 (2022-11-22) Co-DETR (Swin-L) — 64.8 (2022-11-22) GRiT (ViT-H, single-scale testing) — 60.4 (2022-12-01) GRiT (ViT-H, single-scale testing) — 60.4 (2022-12-01) GRiT (ViT-H, single-scale testing) — 60.4 (2022-12-01) GRiT (ViT-H, single-scale testing) — 60.4 (2022-12-01) GRiT (ViT-H, single-scale testing) — 60.4 (2022-12-01) DETA (Swin-L) — 63.5 (2022-12-12) DETA (Swin-L) — 63.5 (2022-12-12) DETA (Swin-L) — 63.5 (2022-12-12) DETA (Swin-L) — 63.5 (2022-12-12) DETA (Swin-L) — 63.5 (2022-12-12) RevCol-H(DINO) — 63.8 (2022-12-22) RevCol-H(DINO) — 63.8 (2022-12-22) RevCol-H(DINO) — 63.8 (2022-12-22) RevCol-H(DINO) — 63.8 (2022-12-22) RevCol-H(DINO) — 63.8 (2022-12-22) Plain-DETR (Swin-L) — 63.9 (2023-01-01) Plain-DETR (Swin-L) — 63.9 (2023-01-01) Plain-DETR (Swin-L) — 63.9 (2023-01-01) Plain-DETR (Swin-L) — 63.9 (2023-01-01) Plain-DETR (Swin-L) — 63.9 (2023-01-01) Grounding DINO — 63.0 (2023-03-09) Grounding DINO — 63.0 (2023-03-09) Grounding DINO — 63.0 (2023-03-09) Grounding DINO — 63.0 (2023-03-09) Grounding DINO — 63.0 (2023-03-09) Focal-Stable-DINO (Focal-Huge, no TTA) — 64.8 (2023-04-25) Focal-Stable-DINO (Focal-Huge, no TTA) — 64.8 (2023-04-25) Focal-Stable-DINO (Focal-Huge, no TTA) — 64.8 (2023-04-25) Focal-Stable-DINO (Focal-Huge, no TTA) — 64.8 (2023-04-25) Focal-Stable-DINO (Focal-Huge, no TTA) — 64.8 (2023-04-25) Swin-S (RPE w/ GAB) — 48.23 (2023-05-08) Swin-S (RPE w/ GAB) — 48.23 (2023-05-08) Swin-S (RPE w/ GAB) — 48.23 (2023-05-08) Swin-S (RPE w/ GAB) — 48.23 (2023-05-08) Swin-S (RPE w/ GAB) — 48.23 (2023-05-08) MoCaE — 65.1 (2023-09-26) MoCaE — 65.1 (2023-09-26) MoCaE — 65.1 (2023-09-26) MoCaE — 65.1 (2023-09-26) MoCaE — 65.1 (2023-09-26) GLEE-Pro — 62.3 (2023-12-14) GLEE-Plus — 60.6 (2023-12-14) GLEE-Lite — 54.7 (2023-12-14) GLEE-Pro — 62.3 (2023-12-14) GLEE-Plus — 60.6 (2023-12-14) GLEE-Lite — 54.7 (2023-12-14) GLEE-Pro — 62.3 (2023-12-14) GLEE-Plus — 60.6 (2023-12-14) GLEE-Lite — 54.7 (2023-12-14) GLEE-Pro — 62.3 (2023-12-14) GLEE-Plus — 60.6 (2023-12-14) GLEE-Lite — 54.7 (2023-12-14) GLEE-Pro — 62.3 (2023-12-14) GLEE-Plus — 60.6 (2023-12-14) GLEE-Lite — 54.7 (2023-12-14) PIIP-H6B (DINO) — 60.0 (2024-06-06) PIIP-H6B (DINO) — 60.0 (2024-06-06) PIIP-H6B (DINO) — 60.0 (2024-06-06) PIIP-H6B (DINO) — 60.0 (2024-06-06) PIIP-H6B (DINO) — 60.0 (2024-06-06) LeYOLO (Large@768) — 41.0 (2024-06-20) LeYOLO (Medium@640) — 39.3 (2024-06-20) LeYOLO (Small@640) — 38.2 (2024-06-20) LeYOLO (Large@768) — 41.0 (2024-06-20) LeYOLO (Medium@640) — 39.3 (2024-06-20) LeYOLO (Small@640) — 38.2 (2024-06-20) LeYOLO (Large@768) — 41.0 (2024-06-20) LeYOLO (Medium@640) — 39.3 (2024-06-20) LeYOLO (Small@640) — 38.2 (2024-06-20) LeYOLO (Large@768) — 41.0 (2024-06-20) LeYOLO (Medium@640) — 39.3 (2024-06-20) LeYOLO (Small@640) — 38.2 (2024-06-20) LeYOLO (Large@768) — 41.0 (2024-06-20) LeYOLO (Medium@640) — 39.3 (2024-06-20) LeYOLO (Small@640) — 38.2 (2024-06-20) Relation-DETR (Focal-L) — 63.5 (2024-07-16) Relation-DETR (Focal-L) — 63.5 (2024-07-16) Relation-DETR (Focal-L) — 63.5 (2024-07-16) Relation-DETR (Focal-L) — 63.5 (2024-07-16) Relation-DETR (Focal-L) — 63.5 (2024-07-16) Faster R-CNN + FPN — 36.2 (2016-12-09) Faster R-CNN + TDM — 36.8 (2016-12-20) DeformConv-R-FCN (Aligned-Inception-ResNet) — 37.5 (2017-03-17) Mask R-CNN (ResNeXt-101-FPN) — 39.8 (2017-03-20) RetinaNet (ResNeXt-101-FPN) — 40.8 (2017-08-07) RefineDet512+ (ResNet-101) — 41.8 (2017-11-18) D-RFCN + SNIP (DPN-98 with flip, multi-scale) — 45.7 (2017-11-22) PANet (ResNeXt-101, multi-scale) — 47.4 (2018-03-05) TridentNet (ResNet-101-Deformable, Image Pyramid) — 48.4 (2019-01-07) NAS-FPN (AmoebaNet-D, learned aug) — 50.7 (2019-06-26) Cascade Mask R-CNN (Triple-ResNeXt152, multi-scale) — 53.3 (2019-09-09) DetectoRS (ResNeXt-101-64x4d, multi-scale) — 55.7 (2020-06-03) YOLOv4-P7 with TTA — 55.8 (2020-11-16) Cascade Eff-B7 NAS-FPN (1280, self-training Copy Paste, single-scale) — 57.3 (2020-12-13) Swin-L (HTC++, multi scale) — 58.7 (2021-03-25) DyHead (Swin-L, multi scale, self-training) — 60.6 (2021-06-15) Soft Teacher + Swin-L (HTC++, multi-scale) — 61.3 (2021-06-16) SwinV2-G (HTC++) — 63.1 (2021-11-18) DINO (Swin-L,multi-scale, TTA) — 63.3 (2022-03-07) FocalNet-H (DINO) — 64.4 (2022-03-22) Group DETR v2 — 64.5 (2022-11-07) InternImage-H (M3I Pre-training) — 65.5 (2022-11-10) Co-DETR — 66.0 (2022-11-22)
RankModel box mAPAP50AP75APSAPMAPLHardware BurdenParams (M) PaperCodeYear
1 Co-DETR 66.0304 DETRs with Collaborative Hybrid Assignments Training open-mmlab/mmdetection · siyuanliii/masa · sense-x/co-detr · +3 2022
2 InternImage-H (M3I Pre-training) 65.52180 InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions opengvlab/internimage · OpenGVLab/M3I-Pretraining · chenller/mmseg-extension 2022
3 M3I Pre-training (InternImage-H) 65.4 Towards All-in-one Pre-training via Maximizing Multi-modal Mutual Information OpenGVLab/M3I-Pretraining 2022
4 MoCaE 65.1 MoCaE: Mixture of Calibrated Experts Significantly Improves Object Detection fiveai/MoCaE 2023
5 Focal-Stable-DINO (Focal-Huge, no TTA) 64.881.771.548.667.678689 A Strong and Reproducible Object Detector with Only Public Datasets microsoft/FocalNet · idea-research/stable-dino · idea-research/stabledino 2023
5 Co-DETR (Swin-L) 64.8218 DETRs with Collaborative Hybrid Assignments Training open-mmlab/mmdetection · siyuanliii/masa · sense-x/co-detr · +3 2022
7 EVA 64.781.971.748.567.777.9 EVA: Exploring the Limits of Masked Visual Representation Learning at Scale rwightman/pytorch-image-models · open-mmlab/mmselfsup · baaivision/eva · +3 2022
8 Group DETR v2 64.581.871.148.467.277.1 Group DETR v2: Strong Object Detector with Encoder-Decoder Pretraining 2022
9 FocalNet-H (DINO) 64.4 Focal Modulation Networks PaddlePaddle/PaddleDetection · keras-team/keras-io · microsoft/FocalNet · +6 2022
10 InternImage-XL 64.3602 InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions opengvlab/internimage · OpenGVLab/M3I-Pretraining · chenller/mmseg-extension 2022
11 FD-SwinV2-G 64.2 Contrastive Learning Rivals Masked Image Modeling in Fine-tuning via Feature Distillation SwinTransformer/Feature-Distillation 2022
12 Plain-DETR (Swin-L) 63.982.170.748.266.876.7228 DETR Does Not Need Multi-Scale or Locality Design impiga/plain-detr 2023
13 RevCol-H(DINO) 63.8 Reversible Column Networks megvii-research/revcol 2022
14 BEiT-3 63.7 Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks microsoft/unilm · lyan62/data-curation 2022
15 Relation-DETR (Focal-L) 63.580.869.147.266.977.0214 Relation DETR: Exploring Explicit Position Relation Prior for Object Detection xiuqhou/relation-detr · xiuqhou/Salience-DETR 2024
15 DETA (Swin-L) 63.580.470.246.166.976.9 NMS Strikes Back jozhang97/deta 2022
17 DINO (Swin-L,multi-scale, TTA) 63.3 DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection IDEA-Research/Grounded-Segment-Anything · PaddlePaddle/PaddleDetection · lucasjinreal/yolov7_d2 · +13 2022
18 SwinV2-G (HTC++) 63.13000 Swin Transformer V2: Scaling Up Capacity and Resolution rwightman/pytorch-image-models · microsoft/Swin-Transformer · PaddlePaddle/PaddleDetection · +20 2021
19 Grounding DINO 63.0 Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection huggingface/transformers · IDEA-Research/Grounded-Segment-Anything · idea-research/groundingdino · +7 2023
20 Florence-CoSwin-H 62.4 Florence: A New Foundation Model for Computer Vision microsoft/unicl · MindCode-4/code-3 2021
1–20 / 1125 다음 → 페이지당 10 20 50 100