Article
Engineering, Mechanical
Yujin Gao, Tim Cain, Patrick Cooper
Summary: This paper discusses the second order Cyclostationarity Analysis (CA) and Detection of Envelope Modulation On Noise (DEMON) for detection and classification of marine vehicles driven by propellers in broadband passive sonar systems. A Coherent Integration and Quadratic Detection Law (CIQDL) is proposed for constructing cyclo-spectra, which outperforms the Quadratic Detection Law (QDL) with a theoretically-derived SNR gain larger than unity. The in-phase characteristic is demonstrated using both simulation and sea-trial data, and the performance of CIQDL and QDL is quantitatively assessed.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Remote Sensing
Ines Barbero-Garcia, Mieke Kuschnerus, Sander Vos, Roderik Lindenbergh
Summary: Efforts are being made to monitor and understand the dynamics of sandy beaches, as they are subject to changes from natural and anthropogenic factors. This study presents a methodology for detecting anthropogenic changes in coastal ecosystems by automatically identifying active bulldozers in beach video data. By using PCA and the YOLO algorithm, the bulldozers in change images can be accurately detected. The correlation between the results of this methodology and 3D data obtained from a laser scanner was computed for validation.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Computer Science, Information Systems
Alok Singh, Thoudam Doren Singh, Sivaji Bandyopadhyay
Summary: This paper presents a dual-stage approach for generating natural language descriptions for videos. The approach addresses the issue of redundancy caused by similar frames in videos by processing selected sets of frames and keyframes. The first stage involves a novel shot boundary detection approach to segment the video and select keyframes and frames. The second stage combines the extracted features with semantic concepts and uses a recurrent neural network for text generation. The proposed approach combines classical and modern computer vision techniques and has been validated on different datasets.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Review
Engineering, Civil
Jorge E. Espinosa, Sergio A. Velastin, John W. Branch
Summary: Motorcycles are vulnerable road users in urban areas, and automatic video processing using Deep Learning theory shows potential in effectively detecting and tracking them. The paper reviews algorithms used for motorcycle detection and tracking, introduces a new dataset, discusses future challenges, and concludes with proposed future work in this evolving area.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2021)
Article
Automation & Control Systems
Yang Liu, Jing Liu, Kun Yang, Bobo Ju, Siao Liu, Yuzheng Wang, Dingkang Yang, Peng Sun, Liang Song
Summary: This study proposes an appearance-motion prototype network (AMP-net) for detecting anomalous events in surveillance videos. By utilizing external memories to record prototype features and introducing temporal attention to enhance the representation of dynamics, the proposed method achieves a delicate balance of effective representation of normal events and accurate detection of anomalies. Experimental results demonstrate that AMP-net achieves performance comparable to state-of-the-art methods on multiple benchmark datasets.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Hardware & Architecture
Ramin Safa, Peyman Bayat, Leila Moghtader
Summary: The paper proposes an approach to predict depression symptoms by analyzing social platform data, achieving high accuracy through analyzing tweets and user profile text. The authors believe that performance improvements can be achieved by limiting the user domain or presence of clinical information.
JOURNAL OF SUPERCOMPUTING
(2022)
Article
Chemistry, Multidisciplinary
Jesus Salido, Vanesa Lomas, Jesus Ruiz-Santaquiteria, Oscar Deniz
Summary: This study focuses on automatic detection of handguns in video surveillance images using three convolutional neural network models. Results show that including pose information can reduce false positives, with the YOLOv3 model trained on dataset including pose information demonstrating the best performance.
APPLIED SCIENCES-BASEL
(2021)
Article
Veterinary Sciences
Franziska Hakansson, Dan Borge Jensen
Summary: Automated monitoring of pigs using computer vision-based methods can detect tail biting events and help farmers reduce health and welfare issues. The CNN-LSTM method shows better generalization on new data compared to the CNN-CNN method.
FRONTIERS IN VETERINARY SCIENCE
(2023)
Article
Computer Science, Information Systems
Hongchun Yuan, Zhenyu Cai, Hui Zhou, Yue Wang, Xiangzhi Chen
Summary: In this study, a prediction-based video anomaly detection approach named TransAnomaly is proposed, which combines U-Net and ViViT to improve anomaly detection performance. By modifying ViViT for video prediction, richer temporal information and more global contexts are captured. Experimental results demonstrate that adding the transformer module enhances anomaly detection performance, and the model outperforms other state-of-the-art prediction-based video anomaly detection methods with proper settings and regularity score calculations.
Article
Chemistry, Multidisciplinary
Mark Bronakowski, Mahmood Al-khassaweneh, Ali Al Bataineh
Summary: Clickbait headlines are misleading titles that aim to attract attention and encourage users to click on a link. These links can contain malware, trojans, and phishing attacks. Clickbaiting is a subtle method used by hackers and scammers. This paper presents a method that utilizes semantic analysis and machine learning to identify clickbait headlines. The proposed models achieved an accuracy of 98% in classifying clickbait headlines, and can serve as a template for developing practical applications to automatically detect clickbait headlines.
APPLIED SCIENCES-BASEL
(2023)
Article
Clinical Neurology
Chad M. Aldridge, Mark M. McDonald, Mattia Wruble, Yan Zhuang, Omar Uribe, Timothy L. McMurry, Iris Lin, Haydon Pitchford, Brett J. Schneider, William A. Dalrymple, Joseph F. Carrera, Sherita Chapman, Bradford B. Worrall, Gustavo K. Rohde, Andrew M. Southerland
Summary: This study demonstrates the potential of a machine learning algorithm using video analysis to detect common signs of stroke. The algorithm shows comparable accuracy and sensitivity to trained paramedics in detecting the presence and laterality of facial weakness in publicly available videos.
FRONTIERS IN NEUROLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Nigel Bosch, Sidney K. D'Mello
Summary: The study reports on the use of facial features to automatically detect mind wandering in both laboratory and classroom settings, achieving above-chance improvements. The results suggest that integrating mind wandering detectors into intelligent interfaces can enhance engagement and learning.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2021)
Article
Automation & Control Systems
Nasir N. Hurrah, Nazir A. Loan, Shabir A. Parah, Javaid A. Sheikh, Khan Muhammad, Antonio Roberto L. de Macedo, Victor Hugo C. de Albuquerque
Summary: The article introduces an efficient algorithm for detecting and correcting forgery in industrial images, which does not require digital signatures or watermarks. By using basic geometric concepts and intensity correlation computation, the algorithm accurately detects rotation angles in industrial images. Experimental results show that the algorithm is highly efficient and preferred for trustworthy media delivery in industrial automation.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Computer Science, Software Engineering
Alonso Pizarro, Silvano F. Dal Sasso, Salvatore Manfreda
Summary: VISION is an open-source software written in MATLAB for video stabilization using automatic feature detection. It has been mainly developed for image velocimetry applications in rivers. The software offers various options for customization based on user needs and intended applications, allowing for the selection of different feature detection algorithms, definition of the percentual value of features for stabilization, choice of geometric transformation type, definition of a region of interest, and real-time visualization of stabilized frames. A case study demonstrated the effectiveness of VISION in reducing velocity errors compared to field measurements. VISION is a user-friendly software suitable for both research and educational purposes.
Article
Radiology, Nuclear Medicine & Medical Imaging
Nasim Sirjani, Shakiba Moradi, Mostafa Ghelich Oghli, Ali Hosseinsabet, Azin Alizadehasl, Mona Yadollahi, Isaac Shiri, Ali Shabanzadeh
Summary: This study utilized the EchoRCNN method to accurately measure cardiac volume and function using neural network technology, including segmentation of the left and right ventricle regions and estimation of key parameters, providing important insights for clinical diagnosis.
INSIGHTS INTO IMAGING
(2022)