TY - GEN
T1 - Complex-Valued Fully Convolutional Network for PolSAR Image Classification with Noisy Labels
AU - Wang, Ningwei
AU - Bi, Haixia
AU - Wang, Xiaotian
AU - Chen, Zhao
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - The process of annotating PolSAR data is highly intricate and demands a proficient understanding of the subject matter. Due to the complexity inherent in this task, it may result in imprecise or erroneous labels being incorporated. The presence of noisy labels inevitably impacts the performance of models in this context, making PolSAR classification a challenging task. This paper proposes a module for correcting noisy labels, which utilizes a CV-CNN with two convolutional layers as its backbone and presents two key contributions: (1) an effective label correction method that leverages the inherent similarities between training samples to repair imprecise or erroneous labels, and (2) a rebalancing loss function that adjusts the weights of different classes to enhance the accuracy of smaller classes. Experimental evaluations on the Flevoland dataset demonstrate the efficacy of our proposed approach.
AB - The process of annotating PolSAR data is highly intricate and demands a proficient understanding of the subject matter. Due to the complexity inherent in this task, it may result in imprecise or erroneous labels being incorporated. The presence of noisy labels inevitably impacts the performance of models in this context, making PolSAR classification a challenging task. This paper proposes a module for correcting noisy labels, which utilizes a CV-CNN with two convolutional layers as its backbone and presents two key contributions: (1) an effective label correction method that leverages the inherent similarities between training samples to repair imprecise or erroneous labels, and (2) a rebalancing loss function that adjusts the weights of different classes to enhance the accuracy of smaller classes. Experimental evaluations on the Flevoland dataset demonstrate the efficacy of our proposed approach.
KW - Complex-Valued Convolutional Neural Network (CV-CNN)
KW - Polarimetric SAR (PolSAR) image classification
KW - correcting noisy labels
KW - rebalancing loss function
UR - https://www.scopus.com/pages/publications/85178334764
U2 - 10.1109/IGARSS52108.2023.10282961
DO - 10.1109/IGARSS52108.2023.10282961
M3 - 会议稿件
AN - SCOPUS:85178334764
T3 - International Geoscience and Remote Sensing Symposium (IGARSS)
SP - 5962
EP - 5965
BT - IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023
Y2 - 16 July 2023 through 21 July 2023
ER -