student_num_detect/configs/hrnet

• High-resolution networks (HRNets) for object detection
• • Introduction
  • Results and Models
  • • Faster R-CNN
    • Mask R-CNN
    • Cascade R-CNN
    • Cascade Mask R-CNN
    • Hybrid Task Cascade (HTC)
    • FCOS

High-resolution networks (HRNets) for object detection

Introduction

@inproceedings{SunXLW19,
  title={Deep High-Resolution Representation Learning for Human Pose Estimation},
  author={Ke Sun and Bin Xiao and Dong Liu and Jingdong Wang},
  booktitle={CVPR},
  year={2019}
}

@article{SunZJCXLMWLW19,
  title={High-Resolution Representations for Labeling Pixels and Regions},
  author={Ke Sun and Yang Zhao and Borui Jiang and Tianheng Cheng and Bin Xiao
  and Dong Liu and Yadong Mu and Xinggang Wang and Wenyu Liu and Jingdong Wang},
  journal   = {CoRR},
  volume    = {abs/1904.04514},
  year={2019}
}

Results and Models

Faster R-CNN

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	box AP	Download
HRNetV2p-W18	pytorch	1x	6.6	13.4	36.9	model | log
HRNetV2p-W18	pytorch	2x
HRNetV2p-W32	pytorch	1x	9.0	12.4	40.2	model | log
HRNetV2p-W32	pytorch	2x	9.0		41.4	model | log
HRNetV2p-W40	pytorch	1x	10.4	10.5	41.2	model | log
HRNetV2p-W40	pytorch	2x	10.4		42.1	model | log

Mask R-CNN

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	box AP	mask AP	Download
HRNetV2p-W18	pytorch	1x	7.0	11.7	37.7	34.2	model | log
HRNetV2p-W18	pytorch	2x	7.0	-	39.8	36.0	model | log
HRNetV2p-W32	pytorch	1x	9.4	11.3	41.2	37.1	model | log
HRNetV2p-W32	pytorch	2x	9.4	-	42.5	37.8	model | log
HRNetV2p-W40	pytorch	1x	10.9		42.1	37.5	model | log
HRNetV2p-W40	pytorch	2x	10.9		42.8	38.2	model | log

Cascade R-CNN

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	box AP	Download
HRNetV2p-W18	pytorch	20e	7.0	11.0	41.2	model | log
HRNetV2p-W32	pytorch	20e	9.4	11.0	43.3	model | log
HRNetV2p-W40	pytorch	20e	10.8		43.8	model | log

Cascade Mask R-CNN

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	box AP	mask AP	Download
HRNetV2p-W18	pytorch	20e	8.5	8.5	41.6	36.4	model | log
HRNetV2p-W32	pytorch	20e		8.3	44.3	38.6	model | log
HRNetV2p-W40	pytorch	20e	12.5		45.1	39.3	model | log

Hybrid Task Cascade (HTC)

Backbone	Style	Lr schd	Mem (GB)	Inf time (fps)	box AP	mask AP	Download
HRNetV2p-W18	pytorch	20e	10.8	4.7	42.8	37.9	model | log
HRNetV2p-W32	pytorch	20e	13.1	4.9	45.4	39.9	model | log
HRNetV2p-W40	pytorch	20e	14.6		46.4	40.8	model | log

FCOS

Backbone	Style	GN	MS train	Lr schd	Mem (GB)	Inf time (fps)	box AP	Download
HRNetV2p-W18	pytorch	Y	N	1x	13.0	12.9	35.1	model | log
HRNetV2p-W18	pytorch	Y	N	2x	13.0	-	37.7	model | log
HRNetV2p-W32	pytorch	Y	N	1x	17.5	12.9	39.2	model | log
HRNetV2p-W32	pytorch	Y	N	2x	17.5	-	40.3	model | log
HRNetV2p-W18	pytorch	Y	Y	2x	13.0	12.9	38.1	model | log
HRNetV2p-W32	pytorch	Y	Y	2x	17.5	12.4	41.8	model | log
HRNetV2p-W48	pytorch	Y	Y	2x	20.3	10.8	42.8	model | log

Note:

• The 28e schedule in HTC indicates decreasing the lr at 24 and 27 epochs, with a total of 28 epochs.
• HRNetV2 ImageNet pretrained models are in HRNets for Image Classification.