Article
Computer Science, Information Systems
Al-Omaisi Asia, Cheng-Zhang Zhu, Sara A. Althubiti, Dalal Al-Alimi, Ya-Long Xiao, Ping-Bo Ouyang, Mohammed A. A. Al-Qaness
Summary: This study uses a deep learning approach with a convolutional neural network to detect and classify diabetic retinopathy. Preprocessing, regularization, and augmentation techniques are used to prepare the dataset. Different residual neural network structures are utilized to achieve accurate classification of DR images, with ResNet-101 showing the highest accuracy.
Article
Computer Science, Artificial Intelligence
Ratko Grbic, Brando Koch
Summary: This paper proposes an algorithm that automatically detects parking slots and classifies them as occupied or vacant solely based on input images. The algorithm is evaluated on publicly available datasets and shows high efficiency in parking slot detection and certain degree of robustness to illegal parking or passing vehicles. The trained classifier achieves high accuracy in parking slot occupancy classification.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Economics
Xizhen Zhou, Mengqi Lv, Yanjie Ji, Shuichao Zhang, Yong Liu
Summary: Parking price adjustment is an effective method for influencing parking choices and addressing shortages. This study investigated the mechanisms behind drivers' choice behaviors and evaluated the operational effectiveness of dynamic pricing schemes in Yinzhou District, Ningbo. The results showed that parking cost, cash rewards, occupancy, and road conditions significantly impact parking behavior. Both strategies effectively balance the utilization of parking space-time resources.
Article
Computer Science, Information Systems
Bobae Kim, Sungbin Im, Geonwook Yoo
Summary: This paper proposes an enhanced optimal EPD method based on CNN for plasma etching process, which outperforms traditional methods in performance.
Article
Computer Science, Information Systems
Hyun Kyu Shin, Si Woon Lee, Goo Pyo Hong, Lee Sael, Sang Hyo Lee, Ha Young Kim
Summary: The paper proposes a deep learning-based image object-identification method for detecting defects in concrete structures that commonly occur in underground parking lots, such as paint peeling, leakage peeling, and leakage traces. The faster region-based convolutional neural network (R-CNN) model was used with a training dataset of 6,281 images to objectively localize and detect surface defects. The study verified the performance of the faster RCNN-based defect-detection algorithm along with its applicability to underground parking lots.
CMC-COMPUTERS MATERIALS & CONTINUA
(2021)
Article
Biochemical Research Methods
Chun-Ling Lin, Kun-Chi Wu
Summary: This study proposes a new method for the diagnosis of diabetic retinopathy using deep learning and image recognition technology. By preprocessing and improving the ResNet-50 model, the accuracy is significantly improved. The results show that the revised ResNet-50 model performs better than other common CNN models, avoiding overfitting, reducing loss, and decreasing fluctuation.
BMC BIOINFORMATICS
(2023)
Article
Computer Science, Information Systems
Quang Huy Bui, Jae Kyu Suhr
Summary: This paper proposes a method to apply transformer-based architectures to parking slot detection tasks, using the DETR architecture and fixed anchor points for improved training time and more suitable representation of parking slots. The proposed method shows comparable detection performance to state-of-the-art CNN-based methods.
Article
Computer Science, Information Systems
Narina Thakur, Eshanika Bhattacharjee, Rachna Jain, Biswaranjan Acharya, Yu-Chen Hu
Summary: The rise in traffic congestion today has led to the need for research and development in parking management systems that can provide real-time indications of parking space occupancy. The main challenge is to develop affordable image-based detection methods to replace expensive sensor-based techniques used indoors. With advancements in computer vision and deep learning, convolutional neural networks were used to develop a robust parking occupancy detection framework.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Mathematical & Computational Biology
Abdullah Khan, Asfandyar Khan, Muneeb Ullah, Muhammad Mansoor Alam, Javed Iqbal Bangash, Mazliham Mohd Suud
Summary: Cancer, especially breast cancer, is a prevalent and challenging disease to detect and classify. This study proposes a new convolutional neural network (CNN) model based on VGGNet for breast cancer diagnosis and classification. By reducing the number of layers, the VGGNet-12 model overcomes the problem of overfitting and achieves better performance compared to other models.
FRONTIERS IN COMPUTATIONAL NEUROSCIENCE
(2022)
Article
Remote Sensing
Shuhui Gong, Jiaxin Qin, Haibo Xu, Rui Cao, Yu Liu, Changfeng Jing, Yuxiu Hao, Yuchen Yang
Summary: The prediction of parking occupancy is crucial for urban planning. Previous studies have not considered the correlation between car parks, leading to low prediction accuracy. To address this issue, this study proposes a Temporal-GCN-based correlated parking prediction model (CPPM) that takes into account temporal parking records, similarities in car parks, and their spatial correlations.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Computer Science, Theory & Methods
Lea Dujic Rodic, Toni Perkovic, Maja Kiljo, Petar Olic
Summary: This paper studies the privacy leakage of LoRaWAN smart parking communication devices. It explores how the variation in signal strength caused by vehicle obstruction can transmit information about parking space occupancy, enabling the implementation of a passive side-channel attack at large distances.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2023)
Article
Thermodynamics
Hsin-Yi Chien, Yu-Chen Wang, Guan-Chen Chen
Summary: The research aims to integrate an image recognition system into robotic arm motion for automatic classification, achieving a recognition accuracy of 95%. By using Keras, Tensorflow, and a five-axis robotic arm controlled by an Arduino Uno board, researchers successfully built a convolutional neural network model and planned the trajectory of the arm based on Denavit-Hartenberg parameters.
ADVANCES IN MECHANICAL ENGINEERING
(2021)
Article
Computer Science, Information Systems
Junyi Zou, Wenbin Guo, Feng Wang
Summary: In this research, a deep learning-based technique is proposed to classify and identify road surfaces using an improved VGGNet-16 model and transfer learning strategy. Experimental results show that the improved VGGNet-16 model achieves an accuracy of 96.87% for the classification and recognition of pavements.
Review
Thermodynamics
Yan Ding, Shuxue Han, Zhe Tian, Jian Yao, Wanyue Chen, Qiang Zhang
Summary: The study focuses on the deviation of energy simulation results for buildings from actual consumption, attributing it mainly to inaccurate estimation of occupancy. Improving the accuracy level of occupancy prediction can notably reduce the error between reality and prediction. Despite various studies on occupancy, differences in detection, prediction, and validation methods still exist.
BUILDING SIMULATION
(2022)
Article
Computer Science, Artificial Intelligence
S. Amutha
Summary: This research presents a new hybrid technique for accurately classifying white blood cell (WBC) leukemia. The image quality is enhanced through Contrast Limited Adaptive Histogram Equalization (CLAHE) preprocessing, and segmentation is done using Hidden Markov Random Fields (HMRF). Features are extracted from WBC images using the powerful Convolutional Neural Network (CNN) architecture Visual Geometry Group Network (VGGNet). An Efficient Salp Swarm Algorithm (ESSA) is then employed to optimize the extracted features. The proposed method achieves high accuracy rates of 98.1% and 98.8% on two Acute Lymphoblastic Leukemia Image Databases, respectively. The combination of CNN feature extraction with ESSA feature optimization has potential applications in various image classification tasks.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2023)
Article
Chemistry, Analytical
Roberto Martinez-Velazquez, Diana P. Tobon, Alejandro Sanchez, Abdulmotaleb El Saddik, Emil Petriu
Summary: The novel coronavirus has caused the COVID-19 pandemic, leading to restrictions on activities and significant economic impact worldwide. Developing countries face more challenges due to limited diagnostic resources. A study introduced a method for detecting COVID-19 infections based on self-reported symptoms, with promising results that support further research efforts on machine learning tests for COVID-19 detection.
Article
Engineering, Electrical & Electronic
Xuezhi Xiang, Zhiyuan Wang, Yulong Qiao
Summary: This paper introduces a deep learning-based pavement crack detection network that effectively and accurately detects pavement cracks using Transformer module and other techniques. It has strong potential for industrial applications.
IEEE SENSORS JOURNAL
(2022)
Review
Computer Science, Artificial Intelligence
Xuezhi Xiang, Rokia Abdein, Ning Lv
Summary: The study highlights the issue of missing or loss of certain details in the flow map estimated by the flow estimation network, resulting in decreased accuracy of optical flow. To address this, a CNN with transformer architecture and an occlusion compensation loss are proposed to enhance the feature representation and improve flow accuracy in occluded regions.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Education & Educational Research
Poppy DesClouds, Natalie Durand-Bush, Michael Del Bel, Fedwa Laamarti, Bradley W. Young, Abdulmotaleb El Saddik
Summary: This exploratory study presents a method to capture data on student-athletes' smartphone usage and psychosocial outcomes. The study found that the relationship between smartphone usage and psychosocial outcomes is complex and nuanced, supporting the need for further research on individual characteristics.
JOURNAL FOR THE STUDY OF SPORTS AND ATHLETES IN EDUCATION
(2022)
Article
Computer Science, Hardware & Architecture
Zijian Long, Haiwei Dong, Abdulmotaleb El Saddik
Summary: In the digital era, XR is considered the next frontier. However, limited computing resources and strict latency constraints pose challenges to XR devices in offering quality user experience. Remote rendering can effectively address this issue and has been found to significantly improve the QoE for the New York City model by at least 21% compared to local rendering.
IEEE CONSUMER ELECTRONICS MAGAZINE
(2022)
Article
Chemistry, Analytical
Izaldein Al-Zyoud, Fedwa Laamarti, Xiaocong Ma, Diana Tobon, Abdulmotaleb El Saddik
Summary: This research proposes a data-driven digital twin system to fuse three human physiological bio-signals: heart rate, breathing rate, and blood oxygen saturation level. With the use of computer vision and machine learning technologies, accurate modeling and measurement of the bio-signals are achieved, showing strong performance compared to the ground-truth values. This study lays the foundation and direction for realizing a holistic human health digital twin model for real-world medical applications.
Article
Imaging Science & Photographic Technology
Haiyun Zhou, Xuezhi Xiang, Yujian Qiu, Xuzhao Liu
Summary: This article presents a new graph convolutional network which combines the advanced decoupling graph convolutional network (DC-GCN) with spatial, temporal, channel (STC) series attention module and adaptive normalization (AN). The STC attention module helps to extract important information from skeleton features. Experimental results show that our method achieves competitive accuracy with state-of-the-art action recognition methods.
IMAGING SCIENCE JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Ning Lv, Xuezhi Xiang, Xinyao Wang, Yulong Qiao, Abdulmotaleb El Saddik
Summary: This research proposes a global-aware and local-aware enhancement network for person search, which learns global context using hierarchical vision transformers and incorporates local context and multi-level features using a local-aware enhancement module. Experimental results demonstrate comparable performance to state-of-the-art methods on two benchmark datasets.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2023)
Article
Engineering, Civil
Yazhou Zhang, Prayag Tiwari, Qian Zheng, Abdulmotaleb El Saddik, M. Shamim Hossain
Summary: Traffic events are a major cause of traffic accidents, and detecting these events poses a challenge in traffic management and intelligent transportation systems (ITSs). This paper proposes a multimodal coupled graph attention network (MCGAT) that extracts valuable information from various traffic data sources and represents it in a graphical structure. The proposed model outperforms state-of-the-art baselines in terms of F1 and accuracy, with significant improvements.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Information Systems
Jiaqi Zhao, Hanzheng Wang, Yong Zhou, Rui Yao, Silin Chen, Abdulmotaleb El Saddik
Summary: This paper proposes a discriminative feature learning network based on a visual Transformer for VI-ReID. By using a spatial feature awareness module and a channel feature enhancement module, the network captures long-term dependencies and improves feature representation. Experimental results demonstrate the significant advantages of the proposed method in VI-ReID tasks.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Geochemistry & Geophysics
Xuezhi Xiang, Dashuai Tian, Ning Lv, Qiannan Yan
Summary: Change detection using deep learning methods still face difficulties in detecting changing target edges and small targets. In this paper, a novel CD method named FCDNet is proposed, which utilizes full-scale skip connections and coordinate attention to address these challenges. By utilizing shallow information and high-level semantics and enhancing local relationships and long-distance dependencies, the proposed method achieves comparable performance to state-of-the-art methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Qingtian Yu, Haopeng Wang, Fedwa Laamarti, Abdulmotaleb El Saddik
Summary: Exercise is a popular topic in modern society, and this paper proposes a novel multitask system that combines human pose estimation, action recognition, and repetitive counting. Utilizing heatmaps as features, the system achieved high accuracy in exercise recognition and repetition counting on the Rep-Penn dataset, surpassing existing frameworks.
MULTIMODAL TECHNOLOGIES AND INTERACTION
(2021)
Article
Computer Science, Information Systems
Haopeng Wang, Diana P. Tobon, M. Shamim Hossain, Abdulmotaleb El Saddik
Summary: This paper introduces an emotion care system based on big data analysis for training autism disorder patients, detecting emotions through facial expressions. The system, utilizing deep learning techniques, can accurately recognize six facial expressions and achieve a high accuracy rate of 95.89%.
Article
Computer Science, Information Systems
Francisco Abad-Navarro, Catalina Martinez-Costa, Jesualdo Tomas Fernandez-Breis
Summary: The Semantic Web technologies make data easily readable by computer agents, enabling the automation of complex tasks and facilitating data integration. The approach of Semankey automatically builds SPARQL queries from user-entered keywords by identifying semantic entities and applying query size-based heuristics. The use of query filters and generation of multiple SPARQL queries based on different interpretations of the input according to a domain ontology are the main contributions of Semankey.
Article
Computer Science, Information Systems
Rahatara Ferdousi, M. Anwar Hossain, Abdulmotaleb El Saddik
Summary: Health CPS, a variant of CPS in the healthcare sector, integrates IoT technology to effectively process health sensor data for early prediction of non-communicable diseases like diabetes.