TY - GEN
T1 - An Active Deep Learning Approach for Minimally-Supervised Polsar Image Classification
AU - Bi, Haixia
AU - Xu, Feng
AU - Wei, Zhiqiang
AU - Han, Yibo
AU - Cui, Yuanlong
AU - Xue, Yong
AU - Xu, Zongben
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/7
Y1 - 2019/7
N2 - Aiming at improving the classification performance with greatly reduced annotation cost, this paper presents an active deep learning approach for minimally-supervised PolSAR image classification, which integrates active learning and fine-tuning convolutional neural network (CNN) into a principled framework. Starting from a CNN trained using a very limited number of labeled pixels, we iteratively and actively select the most informative candidates for annotation, and incrementally fine-tune the CNN by incorporating the newly annotated pixels. Moreover, to boost the performance and robustness of the proposed method, we employ Markov random field to enforce label smoothness, and data augmentation technique to enlarge the training set. Extensive experiments demonstrated that our approach achieved state-of-the-art classification results with significantly reduced annotation cost.
AB - Aiming at improving the classification performance with greatly reduced annotation cost, this paper presents an active deep learning approach for minimally-supervised PolSAR image classification, which integrates active learning and fine-tuning convolutional neural network (CNN) into a principled framework. Starting from a CNN trained using a very limited number of labeled pixels, we iteratively and actively select the most informative candidates for annotation, and incrementally fine-tune the CNN by incorporating the newly annotated pixels. Moreover, to boost the performance and robustness of the proposed method, we employ Markov random field to enforce label smoothness, and data augmentation technique to enlarge the training set. Extensive experiments demonstrated that our approach achieved state-of-the-art classification results with significantly reduced annotation cost.
KW - Markov random field (MRF)
KW - Polarimetric SAR (PolSAR) image classification
KW - active learning
KW - convolutional neural network (CNN)
KW - data augmentation
KW - fine-tuning
UR - https://www.scopus.com/pages/publications/85077700507
U2 - 10.1109/IGARSS.2019.8899214
DO - 10.1109/IGARSS.2019.8899214
M3 - 会议稿件
AN - SCOPUS:85077700507
T3 - International Geoscience and Remote Sensing Symposium (IGARSS)
SP - 3185
EP - 3188
BT - 2019 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2019 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 39th IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2019
Y2 - 28 July 2019 through 2 August 2019
ER -