Article
Chemistry, Analytical
Wen-Sheng Wu, Zhe-Ming Lu
Summary: This paper proposes an automated inventory management system using improved YOLOv3 algorithm, which achieves higher detection FPS and mAP, and reduces the average error rate. The accurately counted number of cups and its change provide significant data for inventory management.
Article
Computer Science, Information Systems
Xinbei Jiang, Tianhan Gao, Zichen Zhu, Yukang Zhao
Summary: The rapid outbreak of COVID-19 has led to a growing demand for automatic real-time mask detection services. This paper proposed the Properly Wearing Masked Face Detection Dataset (PWMFD) and the SE-YOLOv3 mask detector, which outperformed other detectors on PWMFD according to experimental results.
Article
Green & Sustainable Science & Technology
Jye-Hwang Lo, Lee-Kuo Lin, Chu-Chun Hung
Summary: The construction industry is highly dangerous, and effectively managing the usage of personal protective equipment (PPE) is crucial. This study applies deep learning to verify whether construction workers are equipped with PPE according to regulations, aiming to reduce the probability of occupational disasters caused by inappropriate use of PPE. The YOLOv3, YOLOv4, and YOLOv7 algorithms are used to detect helmets and high-visibility vests in real-time images or videos. The model achieved a 97% mean average precision (mAP) and 25 frames per second (FPS) in tests, showing its applicability for real-time PPE detection at construction sites.
Article
Chemistry, Analytical
Addie Ira Borja Parico, Tofael Ahamed
Summary: This study developed a real-time pear fruit counter using YOLOv4 and Deep SORT algorithm, finding a balance between accuracy, speed, and computational cost, as well as providing a method for choosing the most suitable model for agricultural science applications. YOLOv4-CSP was identified as the most accurate model, while YOLOv4-tiny was deemed more suitable for applications requiring lower speed and computational cost.
Article
Engineering, Civil
Xu Han, Lining Zhao, Yue Ning, Jingfeng Hu
Summary: This study used an improved YOLO-V4 detection model (ShipYOLO) to detect ships, with three main improvements compared to YOLO-V4: optimized backbone network, new amplified receptive field module, and improved feature pyramid structure. These improvements enhanced the speed and accuracy of ship detection.
JOURNAL OF ADVANCED TRANSPORTATION
(2021)
Article
Engineering, Multidisciplinary
Yong Wu, Xiumin Chu, Lei Deng, Jinyu Lei, Wei He, Grzegorz Krolczyk, Zhixiong Li
Summary: A multi-sensor fusion perception system is proposed in this study to monitor ship motion in inland waterways. By utilizing target detection and tracking algorithms, track association algorithms, the ship motion data collected from multiple sensors are integrated. The experimental results indicate that the integrated ship motion perception system with multiple sensors can significantly improve the information consistency and data accuracy of ship motion.
Article
Engineering, Electrical & Electronic
Jianwei Li, Jie Chen, Pu Cheng, Zhentao Yu, Lu Yu, Cheng Chi
Summary: Deep learning has played a crucial role in the development of synthetic aperture radar (SAR) ship detection. However, the heavy and computation intensive nature of the detectors hinders their usage on the edge. To overcome this issue, lightweight networks and acceleration ideas have been proposed. This survey reviews real-time SAR ship detection papers, covering model compression, acceleration methods, and various object detection techniques. It provides an overview of 70 papers, including years, datasets, journals, deep-learning frameworks, and hardware, as well as experimental results showing significant improvements in speed and accuracy of the algorithms.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Optics
Sri S. Jamiya, Esther P. Rani
Summary: This study proposes a lightweight deep neural network model based on YOLOv3-tiny network, which effectively detects vehicles in real-time by introducing spatial pyramid pooling technology to increase the accuracy and speed of vehicle detection, especially in complex scenes. The network is trained on datasets including vehicle categories such as car, bus, and truck, achieving high accuracy vehicle detection in real-time video frames and various weather conditions.
Article
Chemistry, Multidisciplinary
Shaojian Song, Yuanchao Li, Qingbao Huang, Gang Li
Summary: The proposed video object detection and recognition method has a wide range of applications in self-driving vehicles, intelligent transportation systems, and video surveillance. By employing data augmentation and improving the algorithm structure, our method has significantly enhanced detection performance and accuracy.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Information Systems
Qiwei Xu, Hong Huang, Chuan Zhou, Xuefeng Zhang
Summary: An improved YOLOv3 network model was proposed for real-time detection of high-voltage lead connector faults, achieving an average detection accuracy of 84.26%. The model enhances feature extraction and recognition performance for small targets by introducing dilated convolution, allowing for real-time detection of faults within 0.308 seconds.
Article
Chemistry, Analytical
Meijing Gao, Yang Bai, Zhilong Li, Shiyu Li, Bozhi Zhang, Qiuyue Chang
Summary: This paper presents a jellyfish detection method based on convolution neural network theory and digital image processing technology, improving the image quality by enhancing the underwater image preprocessing algorithm. By establishing a dataset containing seven species of jellyfishes and fish and using the YOLOv3 algorithm for detection, the improved algorithms have been shown to enhance detection accuracy while ensuring real-time detection speed.
Article
Engineering, Marine
Fumin Wu, Qianqian Chen, Yuanqiao Wen, Changshi Xiao, Feier Zeng
Summary: A deep learning-based multi-sensor hierarchical detection method is proposed in this paper for tracking ship exhaust behavior. The method can accurately and quickly locate the detection area of ship exhaust behavior and achieve real-time detection and tracking.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Imran Ahmed, Misbah Ahmad, Awais Ahmad, Gwanggil Jeon
Summary: 5G technology significantly impacts video surveillance by improving person tracking through a deep learning-based framework that combines YOLOv3 detection and Deep SORT tracking algorithms, with the use of transfer learning to enhance model accuracy. Experimental results demonstrate the effectiveness of transfer learning in person tracking, achieving an overall detection accuracy of 95% with YOLOv3 and a tracking accuracy of 96% with Deep SORT.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2021)
Article
Computer Science, Artificial Intelligence
W. B. Zhang, C. Y. Wu, Z. S. Bao
Summary: This study proposes a SA-BiSeNet model to improve water surface segmentation performance by better fusing different features and enhancing feature representation. Experimental results demonstrate the model's excellent performance in terms of segmentation precision and inference speed.
IET IMAGE PROCESSING
(2023)
Article
Engineering, Marine
Xiaoli Yuan, Chengji Yuan, Wuliu Tian, Gan Liu, Jinfen Zhang
Summary: Path planning is crucial for safe navigation of inland ferries. This study proposes a reinforced deep learning method for path planning of ferries considering both economy and safety. The method is verified to be effective through case studies.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Review
Engineering, Multidisciplinary
Farhan Mumtaz, Nor Zaihar Yahaya, Sheikh Tanzim Meraj, Balbir Singh, Ramani Kannan, Oladimeji Ibrahim
Summary: With the increasing need for energy consumption to meet global demands, hybrid renewable energy generation is being studied as an alternative to conventional methods. While renewable energy sources have drawbacks, such as variability in photovoltaic and wind systems and high costs for fuel cells, researchers are focusing on using power electronic devices like DC-DC converters to address these issues and improve efficiency. The selection and integration of appropriate DC-DC converters, along with control techniques like PID, SMC, MPC, SSM, and FLC, play a crucial role in enhancing the overall performance of power systems.
AIN SHAMS ENGINEERING JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Lashari Ghulam Abbas, Zhou Ai, Farhan Mumtaz, Atta Muhammad, Yutang Dai, Rashda Parveen
Summary: The study presents a hybrid interferometer design that utilizes a combination of different fiber types to achieve simultaneous measurement of strain and temperature. Experimental results demonstrate that the hybrid interferometer has high sensitivity and can be widely applied in various fields.
IEEE SENSORS JOURNAL
(2021)
Article
Chemistry, Analytical
Farhan Mumtaz, Muhammad Roman, Bohong Zhang, Lashari Ghulam Abbas, Muhammad Aqueel Ashraf, Yutang Dai, Jie Huang
Summary: A highly sensitive strain sensor based on tunable cascaded Fabry-Perot interferometers is proposed and experimentally demonstrated. The sensor exhibits optimum strain sensitivity and ultra-low temperature sensitivity, providing good performance for strain sensing.
Article
Engineering, Electrical & Electronic
Farhan Mumtaz, Muhammad Roman, Bohong Zhang, Lashari Ghulam Abbas, Muhammad Aqueel Ashraf, Muhammad Arshad Fiaz, Yutang Dai, Jie Huang
Summary: This work presents a highly sensitive optical fiber Surface Plasmon Resonance (SPR) sensor for simultaneous measurement of dual parameters. The sensor utilizes a D-shaped fiber with a nanolayer coating of Silver (Ag) and Hematite (alpha-Fe2O3), with a Fiber Bragg Grating (FBG) for temperature compensation. Finite Element Method (FEM) modeling is used to analyze the sensor's refractive index and temperature response. The experiment shows that the SPR sensor has excellent performance and is suitable for various dual-parameter sensing applications.
IEEE PHOTONICS JOURNAL
(2022)
Article
Optics
Farhan Mumtaz, Dinesh Reddy Alla, Muhammad Roman, Bohong Zhang, Jeffrey D. Smith, Rex E. Gerald II, Ronald J. O'Malley, Jie Huang
Summary: This research presents an innovative technique that joins silica glass fiber and sapphire single-crystal optical fiber, with the splice demonstrating thermal stability up to 1000 degrees C. A strong and durable splice was achieved using a filament heating process. A femtosecond laser was utilized to create a fiber Bragg gratings sensor in the sapphire single-crystal optical fiber, enabling measurement of high-temperature capabilities and signal attenuation characteristics of the splice joint. The experimental results confirm that the proposed splicing method produces a reliable and stable splice joint that can withstand temperatures up to 1000 degrees C, with only 0.5 dB of signal attenuation. This method allows for the integration of sapphire single-crystal fiber-based sensors in extreme environments encountered in various engineering sectors, leading to enhanced process monitoring, product quality, and production efficiency.
Article
Engineering, Electrical & Electronic
Muhammad Roman, Hanok Tekle, Dinesh Reddy Alla, Farhan Mumtaz, Jeffrey D. Smith, Laura Bartlett, Ronald J. O'Malley, Rex E. Gerald, Jie Huang
Summary: This article explores the prospects of using spatially distributed optical fiber temperature sensors based on Rayleigh OFDR technology in the continuous casting of molten steel. The measurement capability of the optical fiber sensors in a simulated steelmaking environment was demonstrated. The embedded optical fiber sensors were used to measure the temperature distribution in the castable lining during the preheating process of the tundish and during its contact with molten steel.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Optics
Farhan Mumtaz, Ghulam Yaseen, Muhammad Roman, Lashari Ghulam Abbas, Muhammad Aqueel Ashraf, Muhammad Arshad Fiaz, Yutang Dai
Summary: A highly non-linear and ultra-sensitive modified core design of a photonic crystal fiber (PCF) sensor is presented for the detection of liquid analytes. The PCF sensor exhibits high relative sensitivity, strong non-linearity, low confinement loss, and operates in a wide wavelength range. It is a simple design that can be fabricated using existing manufacturing technologies.
JOURNAL OF THE OPTICAL SOCIETY OF AMERICA B-OPTICAL PHYSICS
(2023)
Article
Nanoscience & Nanotechnology
Farhan Mumtaz, Muhammad Roman, Bohong Zhang, Lashari Ghulam Abbas, Yutang Dai, Muhammad Aqueel Ashraf, Muhammad Arshad Fiaz, Amit Kumar
Summary: This paper reports a D-shaped surface plasmon resonance (SPR) refractive index (RI) based biosensor for the earlier detection of various bio-analytes. The D-shaped photonic crystal fiber (PCF) lattice provides an improved cladding pattern for interacting with different bio-analytes. The introduction of a nanoscale MXene layer greatly enhances the sensitivity and performance of the sensor. The proposed SPR sensor exhibits high RI sensitivity and detection resolution, and can detect a wide range of bio-analytes.
PHOTONICS AND NANOSTRUCTURES-FUNDAMENTALS AND APPLICATIONS
(2022)
Article
Materials Science, Multidisciplinary
You Wang, Farhan Mumtaz, Yutang Dai
Summary: This research presents the precise micromachining of silicon dioxide wafers using femtosecond laser. A prediction model for groove processing size is developed and an experiment is conducted to investigate the silicon dioxide groove processing technology. The effects of processing parameters on the processed properties of grooves, heat affected zone, and processed roughness are discussed. A variable defocus processing method is proposed to improve the inclination angle of the groove wall and the quality of the machined groove. Scanning electron microscopy is used for groove morphology analysis, allowing for surface quality improvement and processing parameter optimization.
JOURNAL OF LASER APPLICATIONS
(2023)
Article
Materials Science, Multidisciplinary
You Wang, Farhan Mumtaz, Yutang Dai
Summary: Tungsten oxide (WO3) possesses electrochemical, photo-chromic, and gas-chromic characteristics. The seven-core fiber (SCF) generates an interference signal with super-modes. Utilizing the thermo-optic and thermo-expansion characteristics of SCF with the aid of a Pt-WO3 film, the sensor has high sensitivity to the H2 gas environment. The use of spiral micro grooves created by femtosecond-laser ablation significantly enhances H2 sensitivity and achieves a response and recovery time of less than 90 s.
Article
Optics
Farhan Mumtaz, Bohong Zhang, Ronald j. O'malley, Jie Huang
Summary: This research focuses on the performance analysis and characterization of a fiber Bragg gratings (FBGs) array inscribed by a femtosecond laser in a highly multimode coreless fiber. The study evaluates the array's ability to function as a distributed thermal sensing platform and shows its high temperature sensitivity and stability.
Article
Optics
Farhan Mumtaz, Hanok Tekle, Bohong Zhang, Jeffrey d. Smith, Ronald j. O'malley, Jie Huang
Summary: This study reports the fabrication of large-scale, highly cascaded first-order sapphire optical fiber Bragg gratings (FBGs) using a femtosecond laser-assisted point-by-point inscription method. For the first time, a distributed array of 10 FBGs within highly multimode sapphire crystal fiber is demonstrated, enabled by employing a high power laser technique. These first-order FBGs offer advantages such as enhanced reflectivity, shorter fabrication time, and simplified spectral characteristics, making them easier to interpret compared with high-order FBGs.
Article
Computer Science, Information Systems
Jahedul Islam, Sheikh Tanzim Meraj, Ammar Masaoud, Md. Apel Mahmud, Amril Nazir, Muhammad Ashad Kabir, Md. Moinul Hossain, Farhan Mumtaz
Summary: This study proposes an opposition-based quantum bat algorithm (OQBA) to determine the optimum switching angles for multilevel inverters (MLI), formulated by utilizing the habitual characteristics of bats. The algorithm has advanced learning ability to effectively remove lower-order harmonics from the output voltage of MLI, ultimately improving the output voltage quality and MLI efficiency. The algorithm's performance is evaluated through three different case studies involving 7, 11, and 17-level three-phase MLIs, showing substantial improvement and superiority compared to other available algorithms in harmonics reduction and finding correct solutions.