Yue Fan (PhD Student)

MSc Yue Fan

Adresse
Max-Planck-Institut für Informatik
Saarland Informatics Campus
Campus E1 4
66123 Saarbrücken
Standort
E1 4 - 608
Telefon
+49 681 9325 2138
Fax
+49 681 9325 2099

Personal Information

Publications

Fan, Y., Xian, Y., Losch, M. M., & Schiele, B. (2020). Analyzing the Dependency of ConvNets on Spatial Information. Retrieved from https://arxiv.org/abs/2002.01827
(arXiv: 2002.01827)
Abstract
Intuitively, image classification should profit from using spatial information. Recent work, however, suggests that this might be overrated in standard CNNs. In this paper, we are pushing the envelope and aim to further investigate the reliance on spatial information. We propose spatial shuffling and GAP+FC to destroy spatial information during both training and testing phases. Interestingly, we observe that spatial information can be deleted from later layers with small performance drops, which indicates spatial information at later layers is not necessary for good performance. For example, test accuracy of VGG-16 only drops by 0.03% and 2.66% with spatial information completely removed from the last 30% and 53% layers on CIFAR100, respectively. Evaluation on several object recognition datasets (CIFAR100, Small-ImageNet, ImageNet) with a wide range of CNN architectures (VGG16, ResNet50, ResNet152) shows an overall consistent pattern.
Export
BibTeX
@online{Fan_arXiv2002.01827, TITLE = {Analyzing the Dependency of {ConvNets} on Spatial Information}, AUTHOR = {Fan, Yue and Xian, Yongqin and Losch, Max Maria and Schiele, Bernt}, LANGUAGE = {eng}, URL = {https://arxiv.org/abs/2002.01827}, EPRINT = {2002.01827}, EPRINTTYPE = {arXiv}, YEAR = {2020}, MARGINALMARK = {$\bullet$}, ABSTRACT = {Intuitively, image classification should profit from using spatial information. Recent work, however, suggests that this might be overrated in standard CNNs. In this paper, we are pushing the envelope and aim to further investigate the reliance on spatial information. We propose spatial shuffling and GAP+FC to destroy spatial information during both training and testing phases. Interestingly, we observe that spatial information can be deleted from later layers with small performance drops, which indicates spatial information at later layers is not necessary for good performance. For example, test accuracy of VGG-16 only drops by 0.03% and 2.66% with spatial information completely removed from the last 30% and 53% layers on CIFAR100, respectively. Evaluation on several object recognition datasets (CIFAR100, Small-ImageNet, ImageNet) with a wide range of CNN architectures (VGG16, ResNet50, ResNet152) shows an overall consistent pattern.}, }
Endnote
%0 Report %A Fan, Yue %A Xian, Yongqin %A Losch, Max Maria %A Schiele, Bernt %+ Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society %T Analyzing the Dependency of ConvNets on Spatial Information : %G eng %U http://hdl.handle.net/21.11116/0000-0007-80CB-3 %U https://arxiv.org/abs/2002.01827 %D 2020 %X Intuitively, image classification should profit from using spatial information. Recent work, however, suggests that this might be overrated in standard CNNs. In this paper, we are pushing the envelope and aim to further investigate the reliance on spatial information. We propose spatial shuffling and GAP+FC to destroy spatial information during both training and testing phases. Interestingly, we observe that spatial information can be deleted from later layers with small performance drops, which indicates spatial information at later layers is not necessary for good performance. For example, test accuracy of VGG-16 only drops by 0.03% and 2.66% with spatial information completely removed from the last 30% and 53% layers on CIFAR100, respectively. Evaluation on several object recognition datasets (CIFAR100, Small-ImageNet, ImageNet) with a wide range of CNN architectures (VGG16, ResNet50, ResNet152) shows an overall consistent pattern. %K Computer Science, Computer Vision and Pattern Recognition, cs.CV
Fan, Y. (2019). Analyzing the Dependency of ConvNets on Spatial Information. Universität des Saarlandes, Saarbrücken.
Export
BibTeX
@mastersthesis{FanMaster2019, TITLE = {Analyzing the Dependency of {ConvNets} on Spatial Information}, AUTHOR = {Fan, Yue}, LANGUAGE = {eng}, SCHOOL = {Universit{\"a}t des Saarlandes}, ADDRESS = {Saarbr{\"u}cken}, YEAR = {2019}, MARGINALMARK = {$\bullet$}, DATE = {2019}, }
Endnote
%0 Thesis %A Fan, Yue %+ Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society %T Analyzing the Dependency of ConvNets on Spatial Information : %G eng %U http://hdl.handle.net/21.11116/0000-0007-B435-2 %I Universität des Saarlandes %C Saarbrücken %D 2019 %P 75 p. %V master %9 master