TY - GEN
T1 - JM-Net and cluster-SVM for aerial scene classification
AU - Lu, Xiaoqiang
AU - Yuan, Yuan
AU - Fang, Jie
PY - 2017
Y1 - 2017
N2 - Aerial scene classification, which is a fundamental problem for remote sensing imagery, can automatically label an aerial image with a specific semantic category. Although deep learning has achieved competitive performance for aerial scene classification, training the conventional neural networks with aerial datasets will easily stick in overfitting. Because the aerial datasets only contain a few hundreds or thousands images, meanwhile the conventional networks usually contain millions of parameters to be trained. To address the problem, a novel convolutional neural network named Justify Mentioned Net (JM-Net) is proposed in this paper, which has different size of convolution kernels in same layer and ignores the fully convolution layer, so it has fewer parameters and can be trained well on aerial datasets. Additionally, Cluster-SVM, a strategy to improve the accuracy and speed up the classification is used in the specific task. Finally, our method surpass the state-of-art result on the challenging AID dataset while cost shorter time and used smaller storage space.
AB - Aerial scene classification, which is a fundamental problem for remote sensing imagery, can automatically label an aerial image with a specific semantic category. Although deep learning has achieved competitive performance for aerial scene classification, training the conventional neural networks with aerial datasets will easily stick in overfitting. Because the aerial datasets only contain a few hundreds or thousands images, meanwhile the conventional networks usually contain millions of parameters to be trained. To address the problem, a novel convolutional neural network named Justify Mentioned Net (JM-Net) is proposed in this paper, which has different size of convolution kernels in same layer and ignores the fully convolution layer, so it has fewer parameters and can be trained well on aerial datasets. Additionally, Cluster-SVM, a strategy to improve the accuracy and speed up the classification is used in the specific task. Finally, our method surpass the state-of-art result on the challenging AID dataset while cost shorter time and used smaller storage space.
UR - http://www.scopus.com/inward/record.url?scp=85031944828&partnerID=8YFLogxK
U2 - 10.24963/ijcai.2017/332
DO - 10.24963/ijcai.2017/332
M3 - 会议稿件
AN - SCOPUS:85031944828
T3 - IJCAI International Joint Conference on Artificial Intelligence
SP - 2386
EP - 2392
BT - 26th International Joint Conference on Artificial Intelligence, IJCAI 2017
A2 - Sierra, Carles
PB - International Joint Conferences on Artificial Intelligence
T2 - 26th International Joint Conference on Artificial Intelligence, IJCAI 2017
Y2 - 19 August 2017 through 25 August 2017
ER -