A Dual Attention Encoding Network Using Gradient Profile Loss for Oil Spill Detection Based on SAR Images

Article Geochemistry & Geophysics

DBRANet: Road Extraction by Dual-Branch Encoder and Regional Attention Decoder

Si-Bao Chen et al.

Summary: Although road extraction remains challenging in remote sensing image processing, our proposed improved encoder-decoder method DBRANet demonstrates superior segmentation and connectivity accuracy.

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS (2022)

Add to Collection

Article Geochemistry & Geophysics

Oil Spill Contextual and Boundary-Supervised Detection Network Based on Marine SAR Images

Qiqi Zhu et al.

Summary: The proposed oil spill contextual and boundary-supervised detection network (CBD-Net) extracts refined oil spill regions by fusing multiscale features, addressing the issue of poor detection accuracy for small targets in existing models. CBD-Net outperforms other models in extracting robust and accurate oil spill regions from complex SAR images.

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

Joint semantics and data-driven path representation for knowledge graph reasoning

Guanglin Niu et al.

Summary: This paper proposes a novel joint semantics and data-driven path representation that balances explainability and generalization. The simplified paths are obtained by injecting Horn rules, and the entity converter is designed to reduce the heterogeneity between entities and relations. Experimental results show that the proposed model outperforms other models in link prediction and path query answering tasks.

NEUROCOMPUTING (2022)

Add to Collection

Article Environmental Sciences

Comparison of CNNs and Vision Transformers-Based Hybrid Models Using Gradient Profile Loss for Classification of Oil Spills in SAR Images

Abdul Basit et al.

Summary: The paper introduces a method for oil spill detection using CNN and GP loss function. The proposed method achieves good performance in imbalanced classification tasks and shows a 13% improvement compared to existing techniques.

REMOTE SENSING (2022)

Add to Collection

Article Engineering, Biomedical

A hybrid DenseNet121-UNet model for brain tumor segmentation from MR Images

Necip Cinar et al.

Summary: This paper presents a study on brain tumor segmentation using deep learning methods for MR images. The proposed model, combining the UNet architecture and pre-trained DenseNet121 architecture, achieves good performance on the complex structure of tumors in smaller sub-regions. Experimental results demonstrate the superiority of the proposed approach over other state-of-the-art methods for segmenting whole tumor, core tumor, and enhancing tumor.

BIOMEDICAL SIGNAL PROCESSING AND CONTROL (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

Temporal knowledge graph representation learning with local and global evolutions

Jiasheng Zhang et al.

Summary: In this paper, a new framework for temporal knowledge graph (TKG) representation learning with local and global structure evolutions, named EvoExplore, is proposed. The local structure evolution is defined as the establishment process of relations between entities, and a hierarchical-attention-based temporal point process is proposed to capture the formation process of the graph structure. For global structure evolution, a novel soft modularity parameterized by entity representations is proposed to capture the dynamic community partition of the TKGs. Experimental results show the superiority of EvoExplore compared with baseline methods.

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Article Geochemistry & Geophysics

TransRoadNet: A Novel Road Extraction Method for Remote Sensing Images via Combining High-Level Semantic Feature and Context

Zhigang Yang et al.

Summary: This research proposes a road extraction method that combines high-level semantic features and foreground contextual information. By designing a position attention mechanism, constructing a contextual information extraction module, and providing foreground context information, the method improves the expression ability of road features and captures road contextual information effectively. Experimental results demonstrate the superiority of the proposed method over existing methods.

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS (2022)

Add to Collection

Article Chemistry, Analytical

Energy Load Forecasting Using a Dual-Stage Attention-Based Recurrent Neural Network

Alper Ozcan et al.

Summary: The article proposes a novel deep learning model using Dual-Stage Attention-Based Recurrent Neural Networks to predict electric load consumption, achieving good performance in short-term load forecasting.

SENSORS (2021)

Add to Collection

Article Chemistry, Analytical

A Deep-Learning Framework for the Detection of Oil Spills from SAR Data

Mohamed Shaban et al.

Summary: This paper introduces a two-stage deep-learning framework for the identification of oil spill occurrences from SAR images. The framework shows improved precision and Dice score compared to related work, while addressing the issue of reduced oil spill representation in patches through semantic segmentation.

SENSORS (2021)

Add to Collection

Article Environmental Sciences

Feature Merged Network for Oil Spill Detection Using SAR Images

Yonglei Fan et al.

Summary: This study aims to improve the accuracy of the UNet model in identifying oil spill areas and oil-spill-like areas, and alleviate the overfitting problem of the model by proposing a feature merge network (FMNet) for image segmentation. The proposed model obtains more features, allowing for more accurate decision-making. Compared to UNet, the overall accuracy of the proposed method increased by 1.82% to 61.90%, with the recognition accuracy of oil spill areas and oil-spill-like areas increasing by approximately 3% to 56.33%.

REMOTE SENSING (2021)

Add to Collection

Article Computer Science, Information Systems