TY - GEN
T1 - Aggregated residual transformations for deep neural networks
AU - Xie, Saining
AU - Girshick, Ross
AU - Dollár, Piotr
AU - Tu, Zhuowen
AU - He, Kaiming
N1 - Funding Information:
S.X. and Z.T.’s research was partly supported by NSF IIS-1618477. The authors would like to thank Tsung-Yi Lin and Priya Goyal for valuable discussions.
Funding Information:
S.X. and Z.T.'s research was partly supported by NSF IIS-1618477. The authors would like to thank Tsung-Yi Lin and Priya Goyal for valuable discussions.
Publisher Copyright:
© 2017 IEEE.
PY - 2017/11/6
Y1 - 2017/11/6
N2 - We present a simple, highly modularized network architecture for image classification. Our network is constructed by repeating a building block that aggregates a set of transformations with the same topology. Our simple design results in a homogeneous, multi-branch architecture that has only a few hyper-parameters to set. This strategy exposes a new dimension, which we call “cardinality” (the size of the set of transformations), as an essential factor in addition to the dimensions of depth and width. On the ImageNet-1K dataset, we empirically show that even under the restricted condition of maintaining complexity, increasing cardinality is able to improve classification accuracy. Moreover, increasing cardinality is more effective than going deeper or wider when we increase the capacity. Our models, named ResNeXt, are the foundations of our entry to the ILSVRC 2016 classification task in which we secured 2nd place. We further investigate ResNeXt on an ImageNet-5K set and the COCO detection set, also showing better results than its ResNet counterpart. The code and models are publicly available online1.
AB - We present a simple, highly modularized network architecture for image classification. Our network is constructed by repeating a building block that aggregates a set of transformations with the same topology. Our simple design results in a homogeneous, multi-branch architecture that has only a few hyper-parameters to set. This strategy exposes a new dimension, which we call “cardinality” (the size of the set of transformations), as an essential factor in addition to the dimensions of depth and width. On the ImageNet-1K dataset, we empirically show that even under the restricted condition of maintaining complexity, increasing cardinality is able to improve classification accuracy. Moreover, increasing cardinality is more effective than going deeper or wider when we increase the capacity. Our models, named ResNeXt, are the foundations of our entry to the ILSVRC 2016 classification task in which we secured 2nd place. We further investigate ResNeXt on an ImageNet-5K set and the COCO detection set, also showing better results than its ResNet counterpart. The code and models are publicly available online1.
UR - http://www.scopus.com/inward/record.url?scp=85043777453&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85043777453&partnerID=8YFLogxK
U2 - 10.1109/CVPR.2017.634
DO - 10.1109/CVPR.2017.634
M3 - Conference contribution
AN - SCOPUS:85043777453
T3 - Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017
SP - 5987
EP - 5995
BT - Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017
Y2 - 21 July 2017 through 26 July 2017
ER -