Article
Computer Science, Information Systems
Chen Chen, Yanzhi Ren, Hongbo Liu, Yingying Chen, Hongwei Li
Summary: Location awareness is crucial for mobile devices like smartphones. This study proposes a new active sensing system that utilizes acoustic signatures from smartphones to accurately identify indoor location semantics without requiring any additional infrastructure.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Robotics
Nikhil Gosala, Abhinav Valada
Summary: In this work, the authors propose a new method for directly predicting dense panoptic segmentation maps in the BEV, given a single monocular image in the frontal view. The method utilizes a top-down architecture and incorporates a novel dense transformer module to learn the mapping from the FV to the BEV. Evaluation on two datasets shows that the proposed approach outperforms existing techniques in the PQ metric.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Geochemistry & Geophysics
Yongqiang Mao, Kaiqiang Chen, Liangjin Zhao, Wei Chen, Deke Tang, Wenjie Liu, Zhirui Wang, Wenhui Diao, Xian Sun, Kun Fu
Summary: Building 3-D reconstruction from remote sensing images has various applications and we propose an efficient DSM estimation-driven reconstruction framework, called Building3-D, which reconstructs 3-D building models from a single-view remote sensing image. Our Semantic Flow Field-guided DSM Estimation (SFFDE) network improves upon state-of-the-art methods and achieves impressive results in the 3-D point cloud and model reconstruction process.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Multidisciplinary Sciences
Yunjun Nam, Takayuki Sato, Go Uchida, Ekaterina Malakhova, Shimon Ullman, Manabu Tanifuji
Summary: This study focuses on view-tuned face neurons in the inferior temporal (IT) cortex and suggests that view-invariant face representation is established by combining view sensitive visual features. Different faces can be separated in a space where each axis represents one of these features. The complex ventral visual pathway can be approximated by a shallow network comprised of layers of low-level processing and the detection of particular sets of low-level elements derived from natural image fragments.
SCIENTIFIC REPORTS
(2021)
Article
Computer Science, Artificial Intelligence
Tianling Jiang, Zefan Zhang, Xin Li, Yi Ji, Chunping Liu
Summary: In this work, a novel method called Multi-View Semantic Understanding for Visual Dialog (MVSU) is proposed to resolve the visual coreference resolution problem. The model consists of two main textual processing modules, SRR and CRoT. Experimental results demonstrate that MVSU enhances the ability to understand the semantical information in the VisDial v1.0 dataset.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Neurosciences
Viola Mocz, Maryam Vaziri-Pashkam, Marvin M. Chun, Yaoda Xu
Summary: The study demonstrates a nearly orthogonal representation of object identity and nonidentity features throughout the human ventral visual processing pathway, with these nonidentity features largely untangled from the identity features early in visual processing.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Construction & Building Technology
Zilong Wang, Tianhang Zhang, Xiqiang Wu, Xinyan Huang
Summary: This study explores the real-time prediction of transient fire scenarios using deep learning algorithms and external smoke images, demonstrating the potential of deep learning in determining hidden fire information and its application in smart firefighting.
JOURNAL OF BUILDING ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Jingru Tan, Bo Li, Xin Lu, Yongqiang Yao, Fengwei Yu, Tong He, Wanli Ouyang
Summary: This paper addresses the long-tail distribution problem in real-world applications and identifies imbalanced gradients as the main cause of inferior accuracy in tail categories. The authors propose a gradient-driven training mechanism to tackle this problem, rebalancing the positive/negative gradients dynamically. They also introduce a family of gradient-driven loss functions, called equalization losses, which consistently outperform baseline models across various visual tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Information Systems
Yangyang Li, Wei Zhai, Yang Cao, Zheng-Jun Zha
Summary: The paper proposes a novel Location-free Camouflage Generation Network (LCG-Net) that efficiently generates camouflage images with flexible structure and outperforms state-of-the-art methods in multi-appearance regions. The method has unique advantages and potential applications in camouflage generation.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Engineering, Electrical & Electronic
Jingjia Huang, Wei Yan, Ge Li, Thomas Li, Shan Liu
Summary: This paper proposes a new view-based 3D object recognition method that disentangles information from multi-view images to learn comprehensive descriptors while maintaining robustness to variations in view permutation. The method demonstrates effectiveness and competitive performance on various datasets, with codes to be released soon for replication.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Da-Han Wang, Jianmin Li, Shunzhi Zhu
Summary: This paper proposes a novel multi-view visual-semantic representation method for few-labeled visual recognition, which achieves good recognition results under the condition of a limited number of labeled images.
Article
Environmental Sciences
Aaron E. Maxwell, Michelle S. Bester, Christopher A. Ramezan
Summary: This paper discusses various issues that may reduce the reproducibility and replicability of deep learning in remote sensing, including complexity, variable training and assessment processes, and data leakage. It also offers suggestions for enhancing reproducibility and replicability and highlights challenges faced by researchers.
Article
Robotics
Sourav Garg, Niko Suenderhauf, Michael Milford
Summary: This paper presents a novel system that overcomes the challenges in place recognition, which include limited observable visual content, significant viewpoint changes, and radical appearance changes. By utilizing a hybrid image descriptor, keypoint correspondences, and normalization techniques, the system achieves practical performance where existing state-of-the-art methods fail.
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH
(2022)
Article
Agriculture, Multidisciplinary
Tianhua Li, Meng Sun, Qinghai He, Guanshan Zhang, Guoying Shi, Xiaoming Ding, Sen Lin
Summary: In this paper, a recognition and location algorithm based on the improved YOLOv5s is proposed to meet the requirements of intelligent tomato picking. The CARAFE module is used to optimize the upsampling algorithm, increasing the network's receptive field while maintaining its lightweight. EIoU and Quality Focal Loss are introduced to optimize the loss function, solving the problem of reduced accuracy due to uneven samples. The improved YOLOv5s-CQE model achieves higher mAP and recognition accuracy compared to other lightweight networks.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Computer Science, Information Systems
Peng Gao, Hanlin Zhang, Jia Yu, Jie Lin, Xiaopeng Wang, Ming Yang, Fanyu Kong
Summary: This article addresses the issue of low efficiency in training and recognizing object recognition on hyperspectral remote sensing images on resource-constrained devices, proposing a secure and efficient scheme to outsource the task to untrustworthy cloud servers. The scheme protects the privacy of computation input and output and effectively detects misbehavior of the cloud server with optimal probability 1, ensuring security and efficiency.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Article
Engineering, Civil
Tarek Beji, Bouaza Lafdal, Rabah Mehaddi
Summary: The paper presents experimental measurements of transient profiles of burning rates and liquid temperatures in methanol and heptane pool fires. The results suggest significant enhancement in heat transfer due to convective motion within the liquid. Additionally, a two-zone approach for the thermal structure within the liquid has been proposed and will be implemented in a numerical code.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Walter Klein-Hessling
Summary: Fire simulations and analytical validation approaches are becoming increasingly important in the fire safety assessment of nuclear power plants. The COCOSYS code has been improved to better simulate cable fires and has been validated through experiments. It provides a set of input parameters for different ventilation conditions.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Bing Wu, Yu Meng, Yongzheng Yao, Baiwei Lei, Jingxin Wang, Junjie Zhai
Summary: Underground laneway fires have a throttling effect on ventilation systems and the amount of CO produced by combustion is significantly influenced by ventilation conditions.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Ahmad Hassan, Gilbert Accary, Duncan Sutherland, Khalid Moinuddin
Summary: Numerical simulations were conducted to study laboratory-scale junction fires with a shrub fuel bed under different unidirectional wind conditions on various terrain slopes and junction angles. The results showed that the decrease in junction angle enhanced fire spread, and the impact of wind speed on fire behavior depended on the intermediate values of the junction angle. The driving heat transfer mode in junction fires varied based on the presence of wind and slope conditions.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Samuel Vaux, Emeline Georges, Hugues Pretrel
Summary: This study investigates an elevated pool fire scenario in a confined and forced ventilated compartment through both experimental and numerical methods. The objective is to enhance the understanding of the phenomena and evaluate the Computational Fluid Dynamics (CFD) capabilities in this configuration. The comparison between large-scale fire tests and numerical simulations reveals differences in fuel mass loss rate and thermal stratifications between elevated and ground fires. The study also highlights the limitations of the predictive approach commonly used for ground fires when applied to elevated fires.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Mira von der Linde, Meinald T. Thielsch
Summary: This study investigates the motivation, emotions, and value of laypeople in fire incidents and finds that their active involvement provides support. The findings are helpful for emergency managers.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Hong-Zeng Yu, Xiangyang Zhou
Summary: This study conducted fire tests to evaluate the effectiveness of incorporating nitrogen to enhance the fire extinguishing capability of water mist in an obstructed, naturally-ventilated enclosure. The results showed that adding nitrogen did improve the water mist's fire extinguishing capability, although the extent of improvement varied depending on the fire size, obstruction, and ventilation conditions.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Rosaria Ono, Marcos Vargas Valentim, Fulvio Vittorino
Summary: This study presents the analysis of unimpeded walking speed of primary school children in descending movement in stairs, including the differences on walking speeds on the flights and landings and the impact of anthropo-metrical data on their walking speed. The research aimed to provide data for evacuation modelling that is not influenced by factors like crowds or adults controlling the flow. The findings show that anthropometrical characteristics or age do not significantly affect the children's movement in free walking condition, and the reduction factor on flights for children is higher than commonly adopted in simulation models.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Nicolas Correa, Juan Cuevas, Andres Fuentes, Jose Luis Torero, Pedro Reszka
Summary: The use of mass timber framing in high-rise buildings requires accurate structural fire performance calculations. A thermal model incorporating char oxidation is proposed in this study, and its effectiveness is validated by replicating experimental results. However, the model fails to accurately simulate the behavior during the later stages of surface combustion.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Bouaza Lafdal, Rabah Mehaddi, Pascal Boulet, Elmehdi Koutaiba, Tarek Beji
Summary: This paper presents a series of 48 enclosure fire experiments conducted in a 1.40 m cubic compartment. A new oscillatory combustion regime is discovered, characterized by alternations between well-ventilated and under-ventilated conditions. This finding calls for a better understanding and further analysis of the underlying physics.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Shenggang Fan, Xingyang Zheng, Jiacheng Zheng, Meijing Liu, Daoyang Dong
Summary: Based on experimental data and constitutive models, a simplified constitutive model of AISI304 stainless steel at high temperatures was proposed and analyzed in this study. Finite element analysis results showed that the simplified constitutive model produced acceptable results compared to the constitutive model fitted based on experimental data, indicating its feasibility and accuracy in calculating the fire resistance of specimens.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Julian E. Mendez, Martyn S. McLaggan, David Lange
Summary: The study characterised the fire spread dynamics in a simplified ventilated facade by conducting parametric experiments. The results showed a strong relationship between flame height and heat release rate. It was also found that the time for encapsulation failure and subsequent cladding material core ignition decreased as the cavity width was reduced. The study highlights the importance of considering the interaction of the products used in the facade and its geometry for the design of facade assemblies when accounting for the fire performance of the system.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
A. Galgano, C. Di Blasi
Summary: A sensitivity analysis was conducted on a one-dimensional solid-gas model for the ignition of thick poly (methyl methacrylate). The study found that the kinetics of polymer decomposition and monomer combustion play a crucial role in ignition time, and the properties of the polymer and the decomposition atmosphere should be properly considered.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Thomas Gernay
Summary: This paper provides an overview of the application of the Performance-Based Design approach in structural fire design. The value of the approach is demonstrated through case studies, and the process and role of the structural fire engineer are described. Recent research conducted at Johns Hopkins University, including issues such as the coupling between fire and thermal-structural models, characterization of material behavior at elevated temperature, numerical modeling of structures in fire, probabilistic risk assessment, and cost-benefit analyses, is discussed. The paper concludes with a discussion of challenges and future perspectives for performance-based structural fire design.
FIRE SAFETY JOURNAL
(2024)
Article
Engineering, Civil
Salman Ahmad, Hafiz Usman Ahmed, Asad Ali, Xinyi Yang, Ying Huang, Mingwei Guo, Yihao Ren, Pan Lu
Summary: This study investigates the driving behavior patterns of individuals during historical wildfire events and reveals the influence of driving conditions and wildfires on driving behavior, as well as the differences in driving behavior patterns between rural and urban areas.
FIRE SAFETY JOURNAL
(2024)