Grid R-CNN

open-mmlab / mmdetection

Last updated on Feb 23, 2021
Grid R-CNN (R-101, 2x)
Memory (M) 7000.0
inference time (s/im) 0.07937
File Size 319.23 MB
Training Data MS COCO
Training Resources 8x NVIDIA V100 GPUs
Training Time
Architecture FCN, RPN, Convolution, Dilated Convolution, Sigmoid Activation, ResNet, RoIAlign
lr sched 2x
Memory (M) 7000.0
Backbone Layers 101
inference time (s/im) 0.07937
SHOW MORE
SHOW LESS
Grid R-CNN (R-50, 2x)
Memory (M) 5100.0
inference time (s/im) 0.06667
File Size 246.53 MB
Training Data MS COCO
Training Resources 8x NVIDIA V100 GPUs
Training Time
Architecture FCN, RPN, Convolution, Dilated Convolution, Sigmoid Activation, ResNet, RoIAlign
lr sched 2x
Memory (M) 5100.0
Backbone Layers 50
inference time (s/im) 0.06667
SHOW MORE
SHOW LESS
Grid R-CNN (X-101-32x4d, 2x)
Memory (M) 8300.0
inference time (s/im) 0.09259
File Size 317.94 MB
Training Data MS COCO
Training Resources 8x NVIDIA V100 GPUs
Training Time
Architecture FCN, RPN, ResNeXt, Convolution, Dilated Convolution, Sigmoid Activation, RoIAlign
lr sched 2x
Memory (M) 8300.0
Backbone Layers 101
inference time (s/im) 0.09259
SHOW MORE
SHOW LESS
Grid R-CNN (X-101-64x4d, 2x)
Memory (M) 11300.0
inference time (s/im) 0.12987
File Size 468.02 MB
Training Data MS COCO
Training Resources 8x NVIDIA V100 GPUs
Training Time
Architecture FCN, RPN, ResNeXt, Convolution, Dilated Convolution, Sigmoid Activation, RoIAlign
lr sched 2x
Memory (M) 11300.0
Backbone Layers 101
inference time (s/im) 0.12987
SHOW MORE
SHOW LESS
README.md

Grid R-CNN

Introduction

[ALGORITHM]

@inproceedings{lu2019grid,
  title={Grid r-cnn},
  author={Lu, Xin and Li, Buyu and Yue, Yuxin and Li, Quanquan and Yan, Junjie},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  year={2019}
}

@article{lu2019grid,
  title={Grid R-CNN Plus: Faster and Better},
  author={Lu, Xin and Li, Buyu and Yue, Yuxin and Li, Quanquan and Yan, Junjie},
  journal={arXiv preprint arXiv:1906.05688},
  year={2019}
}

Results and Models

Backbone Lr schd Mem (GB) Inf time (fps) box AP Config Download
R-50 2x 5.1 15.0 40.4 config model | log
R-101 2x 7.0 12.6 41.5 config model | log
X-101-32x4d 2x 8.3 10.8 42.9 config model | log
X-101-64x4d 2x 11.3 7.7 43.0 config model | log

Notes:

  • All models are trained with 8 GPUs instead of 32 GPUs in the original paper.
  • The warming up lasts for 1 epoch and 2x here indicates 25 epochs.

Results

Object Detection on COCO minival

Object Detection on COCO minival
MODEL BOX AP
Grid R-CNN (X-101-64x4d, 2x) 43.0
Grid R-CNN (X-101-32x4d, 2x) 42.9
Grid R-CNN (R-101, 2x) 41.5
Grid R-CNN (R-50, 2x) 40.4
Contact us on: hello@paperswithcode.com . Papers With Code is a free resource with all data licensed under CC-BY-SA.
Terms Data policy Cookies policy