• Corpus ID: 244800748

Multi-Label Noise Robust Collaborative Learning Model for Remote Sensing Image Classification

@inproceedings{Aksoy2020MultiLabelNR,
  title={Multi-Label Noise Robust Collaborative Learning Model for Remote Sensing Image Classification},
  author={Ahmet Aksoy and Mahdyar Ravanbakhsh and Beg{\"u}m Demir},
  year={2020}
}

• A. Aksoy, Mahdyar Ravanbakhsh, B. Demir
• Published 19 December 2020
• Computer Science

The development of accurate methods for multilabel classification (MLC) of remote sensing (RS) images is one of the most important research topics in RS. Methods based on Deep Convolutional Neural Networks (CNNs) have shown strong performance gains in RS MLC problems. However, CNN-based methods usually require a high number of reliable training images annotated by multiple land-cover class labels. Collecting such data is time-consuming and costly. To address this problem, the publicly available thematic products, which can include noisy labels, can be used to annotate RS images with zero-labeling cost. However, multi-label noise (which can be associated with wrong and missing label annotations) can distort the learning process of the MLC algorithm. The detection and correction of label noise are challenging tasks, especially in a multi-label scenario, where each image can be associated with more than one label. To address this problem, we propose a novel noise robust collaborative multi-label learning (RCML) method to alleviate the adverse effects of multi-label noise during the training phase of the CNN model. RCML identifies, ranks and excludes noisy multi-labels in RS images based on three main modules: 1) discrepancy module; 2) group lasso module; and 3) swap module. The discrepancy module ensures that the two networks learn diverse features, while producing the same predictions. The task of the group lasso module is to detect the potentially noisy labels assigned to the multi-labeled training images, while the swap module task is devoted to exchanging the ranking information between two networks. Unlike existing methods that make assumptions about the noise distribution, our proposed RCML does not make any prior assumption about the type of noise in the training set. The experiments conducted on two multi-label RS image archives confirm the robustness of the proposed RCML under extreme multi-label noise rates. Our code is publicly available online: http://www.noisy-labels-in-rs.org