No abstract

This paper proposes a new approach based on an unsupervised deep learning (DL) model for landslide detection. Recently, supervised DL models using convolutional neural networks (CNN) have been widely studied for landslide detection. Even though these models provide robust performance and reliable results, they depend highly on a large labeled dataset for their training step. As an alternative, in this paper, we developed an unsupervised learning model by employing a convolutional auto-encoder (CAE) to deal with the problem of limited labeled data for training. The CAE was used to learn and extract the abstract and high-level features without using training data. To assess the performance of the proposed approach, we used Sentinel-2 imagery and a digital elevation model (DEM) to map landslides in three different case studies in India, China, and Taiwan. Using minimum noise fraction (MNF) transformation, we reduced the multispectral dimension to three features containing more than 80% of scene information. Next, these features were stacked with slope data and NDVI as inputs to the CAE model. The Huber reconstruction loss was used to evaluate the inputs. We achieved reconstruction losses ranging from 0.10 to 0.147 for the MNF features, slope, and NDVI stack for all three study areas. The mini-batch K-means clustering method was used to cluster the features into two to five classes. To evaluate the impact of deep features on landslide detection, we first clustered a stack of MNF features, slope, and NDVI, then the same ones plus with the deep features. For all cases, clustering based on deep features provided the highest precision, recall, F1-score, and mean intersection over the union in landslide detection.

High-resolution urban image clustering has remained a challenging task. This is mainly because its performance strongly depends on the discrimination power of features. Recently, several studies focused on unsupervised learning methods by autoencoders to learn and extract more efficient features for clustering purposes. This paper proposes a Boosted Convolutional AutoEncoder (BCAE) method based on feature learning for efficient urban image clustering. The proposed method was applied to multi-sensor remote-sensing images through a multistep workflow. The optical data were first preprocessed by applying a Minimum Noise Fraction (MNF) transformation. Then, these MNF features, in addition to the normalized Digital Surface Model (nDSM) and vegetation indexes such as Normalized Difference Vegetation Index (NDVI) and Excess Green (ExG(2)), were used as the inputs of the BCAE model. Next, our proposed convolutional autoencoder was trained to automatically encode upgraded features and boost the hand-crafted features for producing more clustering-friendly ones. Then, we employed the Mini Batch K-Means algorithm to cluster deep features. Finally, the comparative feature sets were manually designed in three modes to prove the efficiency of the proposed method in extracting compelling features. Experiments on three datasets show the efficiency of BCAE for feature learning. According to the experimental results, by applying the proposed method, the ultimate features become more suitable for clustering, and spatial correlation among the pixels in the feature learning process is also considered.

No abstract