Detectron-Cascade-RCNN

Introduction

This repository re-implements Cascade R-CNN on the base of Detectron. Very consistent gains are available for all tested models, regardless of baseline strength.

Please follow Detectron on how to install and use Detectron-Cascade-RCNN.

It is also recommended to use our original implementation, cascade-rcnn based on Caffe, and the third-party implementation, mmdetection based on PyTorch and tensorpack based on TensorFlow.

Citation

If you use our code/model/data, please cite our paper:

@inproceedings{cai18cascadercnn, author = {Zhaowei Cai and Nuno Vasconcelos}, Title = {Cascade R-CNN: Delving into High Quality Object Detection}, booktitle = {CVPR}, Year = {2018} }

or its extension:

@article{cai2019cascadercnn, author = {Zhaowei Cai and Nuno Vasconcelos}, title = {Cascade R-CNN: High Quality Object Detection and Instance Segmentation}, journal = {arXiv preprint arXiv:1906.09756}, year = {2019} }

and Detectron:

@misc{Detectron2018, author = {Ross Girshick and Ilija Radosavovic and Georgia Gkioxari and Piotr Doll'{a}r and Kaiming He}, title = {Detectron}, howpublished = {url{https://github.com/facebookresearch/detectron}}, year = {2018} }

Benchmarking

End-to-End Faster & Mask R-CNN Baselines

^_backbone	^_type	^_lr schd	^_im/ gpu	^_box AP	^_box AP50	^_box AP75	^_mask AP	^_mask AP50	^_mask AP75	^{_{download links}}
^{_{R-50-FPN-baseline}}	^_Faster	^_1x	^₂	^_36.7	^_58.4	^_39.6	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-50-FPN-cascade}}	^_Faster	^_1x	^₂	^_40.9	^_59.0	^_44.6	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-101-FPN-baseline}}	^_Faster	^_1x	^₂	^_39.4	^_61.2	^_43.4	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-101-FPN-cascade}}	^_Faster	^_1x	^₂	^_42.8	^_61.4	^_46.1	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{X-101-64x4d-FPN-baseline}}	^_Faster	^_1x	^₁	^_41.5	^_63.8	^_44.9	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{X-101-64x4d-FPN-cascade}}	^_Faster	^_1x	^₁	^_45.4	^_64.0	^_49.8	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{X-101-32x8d-FPN-baseline}}	^_Faster	^_1x	^₁	^_41.3	^_63.7	^_44.7	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{X-101-32x8d-FPN-cascade}}	^_Faster	^_1x	^₁	^_44.7	^_63.7	^_48.8	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-50-FPN-baseline}}	^_Mask	^_1x	^₂	^_37.7	^_59.2	^_40.9	^_33.9	^_55.8	^_35.8	^{_{model \| boxes \| masks}}
^{_{R-50-FPN-cascade}}	^_Mask	^_1x	^₂	^_41.3	^_59.6	^_44.9	^_35.4	^_56.2	^_37.8	^{_{model \| boxes \| masks}}
^{_{R-101-FPN-baseline}}	^_Mask	^_1x	^₂	^_40.0	^_61.8	^_43.7	^_35.9	^_58.3	^_38.0	^{_{model \| boxes \| masks}}
^{_{R-101-FPN-cascade}}	^_Mask	^_1x	^₂	^_43.3	^_61.7	^_47.2	^_37.1	^_58.6	^_39.8	^{_{model \| boxes \| masks}}
^{_{X-101-64x4d-FPN-baseline}}	^_Mask	^_1x	^₁	^_42.4	^_64.3	^_46.4	^_37.5	^_60.6	^_39.9	^{_{model \| boxes \| masks}}
^{_{X-101-64x4d-FPN-cascade}}	^_Mask	^_1x	^₁	^_45.9	^_64.4	^_50.2	^_38.8	^_61.3	^_41.6	^{_{model \| boxes \| masks}}
^{_{X-101-32x8d-FPN-baseline}}	^_Mask	^_1x	^₁	^_42.1	^_64.1	^_45.9	^_37.3	^_60.3	^_39.5	^{_{model \| boxes \| masks}}
^{_{X-101-32x8d-FPN-cascade}}	^_Mask	^_1x	^₁	^_45.8	^_64.1	^_50.3	^_38.6	^_60.6	^_41.5	^{_{model \| boxes \| masks}}

^_backbone

^_type

^_lr
schd

^_im/
gpu

^_box
AP

^_box
AP50

^_box
AP75

^_mask
AP

^_mask
AP50

^_mask
AP75

^{_{download
links}}

^{_{R-50-FPN-baseline}}

^_Faster

^_1x

^₂

^_36.7

^_58.4

^_39.6

^_-

^{_{model | boxes}}

^{_{R-50-FPN-cascade}}

^_Faster

^_1x

^₂

^_40.9

^_59.0

^_44.6

^_-

^{_{model | boxes}}

^{_{R-101-FPN-baseline}}

^_Faster

^_1x

^₂

^_39.4

^_61.2

^_43.4

^_-

^{_{model | boxes}}

^{_{R-101-FPN-cascade}}

^_Faster

^_1x

^₂

^_42.8

^_61.4

^_46.1

^_-

^{_{model | boxes}}

^{_{X-101-64x4d-FPN-baseline}}

^_Faster

^_1x

^₁

^_41.5

^_63.8

^_44.9

^_-

^{_{model | boxes}}

^{_{X-101-64x4d-FPN-cascade}}

^_Faster

^_1x

^₁

^_45.4

^_64.0

^_49.8

^_-

^{_{model | boxes}}

^{_{X-101-32x8d-FPN-baseline}}

^_Faster

^_1x

^₁

^_41.3

^_63.7

^_44.7

^_-

^{_{model | boxes}}

^{_{X-101-32x8d-FPN-cascade}}

^_Faster

^_1x

^₁

^_44.7

^_63.7

^_48.8

^_-

^{_{model | boxes}}

^{_{R-50-FPN-baseline}}

^_Mask

^_1x

^₂

^_37.7

^_59.2

^_40.9

^_33.9

^_55.8

^_35.8

^{_{model | boxes | masks}}

^{_{R-50-FPN-cascade}}

^_Mask

^_1x

^₂

^_41.3

^_59.6

^_44.9

^_35.4

^_56.2

^_37.8

^{_{model | boxes | masks}}

^{_{R-101-FPN-baseline}}

^_Mask

^_1x

^₂

^_40.0

^_61.8

^_43.7

^_35.9

^_58.3

^_38.0

^{_{model | boxes | masks}}

^{_{R-101-FPN-cascade}}

^_Mask

^_1x

^₂

^_43.3

^_61.7

^_47.2

^_37.1

^_58.6

^_39.8

^{_{model | boxes | masks}}

^{_{X-101-64x4d-FPN-baseline}}

^_Mask

^_1x

^₁

^_42.4

^_64.3

^_46.4

^_37.5

^_60.6

^_39.9

^{_{model | boxes | masks}}

^{_{X-101-64x4d-FPN-cascade}}

^_Mask

^_1x

^₁

^_45.9

^_64.4

^_50.2

^_38.8

^_61.3

^_41.6

^{_{model | boxes | masks}}

^{_{X-101-32x8d-FPN-baseline}}

^_Mask

^_1x

^₁

^_42.1

^_64.1

^_45.9

^_37.3

^_60.3

^_39.5

^{_{model | boxes | masks}}

^{_{X-101-32x8d-FPN-cascade}}

^_Mask

^_1x

^₁

^_45.8

^_64.1

^_50.3

^_38.6

^_60.6

^_41.5

^{_{model | boxes | masks}}

Mask R-CNN with Bells & Whistles

^_backbone	^_type	^_lr schd	^_im/ gpu	^_box AP	^_box AP50	^_box AP75	^_mask AP	^_mask AP50	^_mask AP75	^{_{download links}}
^{_{X-152-32x8d-FPN-IN5k-baseline}}	^_Mask	^_s1x	^₁	^_48.1	^_68.3	^_52.9	^_41.5	^_65.1	^_44.7	^{_{model \| boxes \| masks}}
^{_{[above without test-time aug.]}}				^_45.2	^_66.9	^_49.7	^_39.7	^_63.5	^_42.4
^{_{X-152-32x8d-FPN-IN5k-cascade}}	^_Mask	^_s1x	^₁	^_50.2	^_68.2	^_55.0	^_42.3	^_65.4	^_45.8	^{_{model \| boxes \| masks}}
^{_{[above without test-time aug.]}}				^_48.1	^_66.7	^_52.6	^_40.7	^_63.7	^_43.8

^_backbone

^_type

^_lr
schd

^_im/
gpu

^_box
AP

^_box
AP50

^_box
AP75

^_mask
AP

^_mask
AP50

^_mask
AP75

^{_{download
links}}

^{_{X-152-32x8d-FPN-IN5k-baseline}}

^_Mask

^_s1x

^₁

^_48.1

^_68.3

^_52.9

^_41.5

^_65.1

^_44.7

^{_{model | boxes | masks}}

^{_{[above without test-time aug.]}}

^_45.2

^_66.9

^_49.7

^_39.7

^_63.5

^_42.4

^{_{X-152-32x8d-FPN-IN5k-cascade}}

^_Mask

^_s1x

^₁

^_50.2

^_68.2

^_55.0

^_42.3

^_65.4

^_45.8

^{_{model | boxes | masks}}

^{_{[above without test-time aug.]}}

^_48.1

^_66.7

^_52.6

^_40.7

^_63.7

^_43.8

Faster & Mask R-CNN with GN

^_backbone	^_type	^_lr schd	^_im/ gpu	^_box AP	^_box AP50	^_box AP75	^_mask AP	^_mask AP50	^_mask AP75	^{_{download links}}
^{_{R-50-FPN-GN-baseline}}	^_Faster	^_1x	^₂	^_38.4	^_59.9	^_41.7	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-50-FPN-GN-cascade}}	^_Faster	^_1x	^₂	^_42.2	^_60.6	^_45.8	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-101-FPN-GN-baseline}}	^_Faster	^_1x	^₂	^_39.9	^_61.3	^_43.3	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-101-FPN-GN-cascade}}	^_Faster	^_1x	^₁	^_43.8	^_62.2	^_47.6	^_-	^_-	^_-	^{_{model \| boxes}}
^{_{R-50-FPN-GN-baseline}}	^_Mask	^_1x	^₂	^_39.2	^_60.5	^_42.9	^_34.9	^_57.1	^_36.9	^{_{model \| boxes}}
^{_{R-50-FPN-GN-cascade}}	^_Mask	^_1x	^₁	^_42.9	^_60.7	^_46.6	^_36.6	^_57.7	^_39.2	^{_{model \| boxes \| masks}}
^{_{R-101-FPN-GN-baseline}}	^_Mask	^_1x	^₂	^_41.1	^_62.1	^_45.1	^_36.3	^_58.9	^_38.5	^{_{model \| boxes \| masks}}
^{_{R-101-FPN-GN-cascade}}	^_Mask	^_1x	^₁	^_44.8	^_62.8	^_48.8	^_38.0	^_59.8	^_40.8	^{_{model \| boxes \| masks}}

^_backbone

^_type

^_lr
schd

^_im/
gpu

^_box
AP

^_box
AP50

^_box
AP75

^_mask
AP

^_mask
AP50

^_mask
AP75

^{_{download
links}}

^{_{R-50-FPN-GN-baseline}}

^_Faster

^_1x

^₂

^_38.4

^_59.9

^_41.7

^_-

^{_{model | boxes}}

^{_{R-50-FPN-GN-cascade}}

^_Faster

^_1x

^₂

^_42.2

^_60.6

^_45.8

^_-

^{_{model | boxes}}

^{_{R-101-FPN-GN-baseline}}

^_Faster

^_1x

^₂

^_39.9

^_61.3

^_43.3

^_-

^{_{model | boxes}}

^{_{R-101-FPN-GN-cascade}}

^_Faster

^_1x

^₁

^_43.8

^_62.2

^_47.6

^_-

^{_{model | boxes}}

^{_{R-50-FPN-GN-baseline}}

^_Mask

^_1x

^₂

^_39.2

^_60.5

^_42.9

^_34.9

^_57.1

^_36.9

^{_{model | boxes}}

^{_{R-50-FPN-GN-cascade}}

^_Mask

^_1x

^₁

^_42.9

^_60.7

^_46.6

^_36.6

^_57.7

^_39.2

^{_{model | boxes | masks}}

^{_{R-101-FPN-GN-baseline}}

^_Mask

^_1x

^₂

^_41.1

^_62.1

^_45.1

^_36.3

^_58.9

^_38.5

^{_{model | boxes | masks}}

^{_{R-101-FPN-GN-cascade}}

^_Mask

^_1x

^₁

^_44.8

^_62.8

^_48.8

^_38.0

^_59.8

^_40.8

^{_{model | boxes | masks}}

智能在线

400-630-6780
聆听.建议反馈

E-mail: support@tusaishared.com