Object segmentation of remotely-sensed aerial (or very-high resolution, VHS) images and satellite (or high-resolution, HR) images, has been applied to many application domains, especially in road extraction in which the segmented objects are served as a mandatory layer in geospatial databases. Several attempts at applying the deep convolutional neural network (DCNN) to extract roads from remote sensing images have been made; however, the accuracy is still limited. In this paper, we present an enhanced DCNN framework specifically tailored for road extraction of remote sensing images by applying landscape metrics (LMs) and conditional random fields (CRFs). To improve the DCNN, a modern activation function called the exponential linear unit (ELU), is employed in our network, resulting in a higher number of, and yet more accurate, extracted roads. To further reduce falsely classified road objects, a solution based on an adoption of LMs is proposed. Finally, to sharpen the extracted roads, a CRF method is added to our framework. The experiments were conducted on Massachusetts road aerial imagery as well as the Thailand Earth Observation System (THEOS) satellite imagery data sets. The results showed that our proposed framework outperformed Segnet, a state-of-the-art object segmentation technique, on any kinds of remote sensing imagery, in most of the cases in terms of precision, recall, and .
In the remote sensing domain, it is crucial to complete semantic segmentation on the raster images, e.g., river, building, forest, etc., on raster images. A deep convolutional encoder–decoder (DCED) network is the state-of-the-art semantic segmentation method for remotely sensed images. However, the accuracy is still limited, since the network is not designed for remotely sensed images and the training data in this domain is deficient. In this paper, we aim to propose a novel CNN for semantic segmentation particularly for remote sensing corpora with three main contributions. First, we propose applying a recent CNN called a global convolutional network (GCN), since it can capture different resolutions by extracting multi-scale features from different stages of the network. Additionally, we further enhance the network by improving its backbone using larger numbers of layers, which is suitable for medium resolution remotely sensed images. Second, “channel attention” is presented in our network in order to select the most discriminative filters (features). Third, “domain-specific transfer learning” is introduced to alleviate the scarcity issue by utilizing other remotely sensed corpora with different resolutions as pre-trained data. The experiment was then conducted on two given datasets: (i) medium resolution data collected from Landsat-8 satellite and (ii) very high resolution data called the ISPRS Vaihingen Challenge Dataset. The results show that our networks outperformed DCED in terms of for 17.48% and 2.49% on medium and very high resolution corpora, respectively.
Sentiment analysis is very important for social listening, especially, when there are millions of Twitter users in Thailand nowadays. Almost all prior works are based on classical classification techniques, e.g., SVM, Naïve Bayes, etc. Recently, the deep learning techniques have shown promising accuracy in this domain on English tweet corpus. In this paper, we propose the first study that applies deep learning techniques to classify sentiment of Thai Twitter data. There are two deep learning techniques included in our study: Long Short Term Memory (LSTM) and Dynamic Convolutional Neural Network (DCNN). A proper data preprocessing has been conducted. Moreover, we also investigate an effect of word orders in Thai tweets. The results show that the deep learning techniques significantly outperform many classical techniques: Naïve Bayes and SVM, except Maximum Entropy.
Transformers have demonstrated remarkable accomplishments in several natural language processing (NLP) tasks as well as image processing tasks. Herein, we present a deep-learning (DL) model that is capable of improving the semantic segmentation network in two ways. First, utilizing the pre-training Swin Transformer (SwinTF) under Vision Transformer (ViT) as a backbone, the model weights downstream tasks by joining task layers upon the pretrained encoder. Secondly, decoder designs are applied to our DL network with three decoder designs, U-Net, pyramid scene parsing (PSP) network, and feature pyramid network (FPN), to perform pixel-level segmentation. The results are compared with other image labeling state of the art (SOTA) methods, such as global convolutional network (GCN) and ViT. Extensive experiments show that our Swin Transformer (SwinTF) with decoder designs reached a new state of the art on the Thailand Isan Landsat-8 corpus (89.8% score), Thailand North Landsat-8 corpus (63.12% score), and competitive results on ISPRS Vaihingen. Moreover, both our best-proposed methods (SwinTF-PSP and SwinTF-FPN) even outperformed SwinTF with supervised pre-training ViT on the ImageNet-1K in the Thailand, Landsat-8, and ISPRS Vaihingen corpora.
Recently, ability to handle tremendous amounts of information using increased computational capabilities has improved prediction of stock market behavior. Complex machine learning algorithms such as deep learning methods can analyze and detect complex data patterns. The recent prediction models use two types of inputs as (i) numerical information such as historical prices and technical indicators, and (ii) textual information including news contents or headlines. However, the use of textual data involves text representation construction. Traditional methods like word embedding may not be suitable for representing the semantics of financial news due to problems of word sparsity in datasets. In this paper, we aim to improve stock market predictions using a deep learning approach with event embedding vectors extracted from news headlines, historical price data, and a set of technical indicators as input. Our prediction model consists of Convolutional Neural Network (CNN) and Long Short-term Memory (LSTM) architectures. We use accuracy and annualized return based on trading simulation as performance metrics, and then perform experiments on three datasets obtained from different news sources namely Reuters, Reddit, and Intrinio. Results show that enhancing text representation vectors and considering both numerical and textual information as input to a deep neural network can improve prediction performance.