Article
Computer Science, Artificial Intelligence
Hanyang Lin, Yongzhao Zhan, Shiqin Liu, Xiao Ke, Yuzhong Chen
Summary: With the widespread use of mobile Internet, mobile payment has become an integral part of daily life. This article proposes an effective method based on deep learning to detect and recognize bank cards in complex natural scenes, achieving accurate recognition of different types of bank cards.
APPLIED INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Xiaofei Bian, Haiwei Pan, Kejia Zhang, Pengyuan Li, Jinbao Li, Chunling Chen
Summary: In this paper, a two-phase classification method for skin lesion images in Asians is proposed, which integrates medical domain knowledge, deep learning, and a refined strategy. A skin-dependent feature is introduced to efficiently distinguish malignant melanoma, and a classification method based on deep learning is proposed. The proposed method significantly improves the classification accuracy of skin diseases compared to state-of-the-art methods.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Chemistry, Analytical
Shuhua Liu, Huixin Xu, Qi Li, Fei Zhang, Kun Hou
Summary: This paper presents an object recognition method based on scene text reading, which improves text detection accuracy and recognition accuracy through deep learning models and dataset training, effectively addressing the issue of robot object recognition in complex scenes.
Article
Engineering, Multidisciplinary
Liqun Hou, Sen Wang, Xiaopeng Sun, Guopeng Mao
Summary: A novel approach based on YOLOX convolutional network and semantic segmentation technology is proposed to improve the accuracy and robustness of reading recognition algorithms for pointer meters. The approach detects the dial of the target meter using the YOLOX network, determines the meter's main tick marks, dial center, and pointer through semantic segmentation, corrects the tilt through perspective transformation, and calculates the reading value using the angles between the pointer and main tick marks. Experimental results show that the presented approach achieves reading values with fiducial errors of no more than 0.31%.
Article
Engineering, Electrical & Electronic
Shiniu Sun, Lisheng Han, Jie Wei, Huimin Hao, Jiahai Huang, Wenbin Xin, Xu Zhou, Peng Kang
Summary: Recognizing sign language quickly and accurately on embedded platforms and mobile terminals is important for meeting the communication needs of hearing impaired individuals and the general public. A lightweight model called ShuffleNetv2-YOLOv3 was proposed for static sign language recognition, which achieved a good balance between accuracy and speed. The model utilized ShuffleNetv2 as the backbone network for YOLOv3, significantly improving the recognition speed. The evaluation of the model's performance showed high F1 score and mAP values of 99.1% and 98.4%, respectively. The mobile terminal application of the lightweight model demonstrated improved inference speeds compared to the original YOLOv3 model. Overall, the ShuffleNetv2-YOLOv3 lightweight model lays a solid foundation for real-time gesture recognition on embedded platforms and mobile terminals.
SIGNAL IMAGE AND VIDEO PROCESSING
(2023)
Article
Computer Science, Information Systems
Riktim Mondal, Samir Malakar, Elisa H. Barney Smith, Ram Sarkar
Summary: Handwriting recognition has been a challenging task, usually requiring large datasets and complex lexicon-based approaches. This study proposes a lexicon-free handwriting recognition technique that is trained on only 1200 word images, achieving successful recognition of handwritten English text without dependency on writers' styles.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Hardware & Architecture
Tingting Wang, Jianqing Li, Wei Wei, Wei Wang, Kai Fang
Summary: This study employs a deep learning-based method for WEMI intrusion detection, extracting sensor fingerprints using Kalman and moving average filters, and extracting frequency and time domain features. Experimental results show that this method performs well in WEMI intrusion detection.
Article
Energy & Fuels
Deyong Li, Huaiwei Ren, Guofa Wang, Shuang Wang, Wenshan Wang, Ming Du
Summary: A real-time detection method for coal gangue based on a multiscale fusion lightweight network (SMS-YOLOv3) is proposed to solve the problems of large memory footprint, low detection speed, and low detection accuracy for small and overlapping targets in the current coal gangue target detection algorithm. The proposed method uses MobileNetv3 as a feature extraction network and adds a shallow detection scale to improve the accuracy of small target detection. Experimental results show that the proposed algorithm achieves accurate and fast detection of small and overlapping targets of coal gangue with an mAP of 98.97%. The algorithm also demonstrates improvements in mAP and fps compared to the original YOLOv3, with a significantly reduced memory footprint.
ENERGY SCIENCE & ENGINEERING
(2023)
Article
Automation & Control Systems
Xiaokang Wang, Laurence Tianruo Yang, Liwen Song, Huihui Wang, Lei Ren, Jamal Deen
Summary: The Industrial Internet-of-Things has transformed industrial manufacturing by incorporating production equipment, mobile terminals, and smart devices with networks, and the processing and recognition methods of industrial visual information are crucial for providing industrial intelligence.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Computer Science, Information Systems
Farman Ali, Sadia Khan, Arbab Waseem Abbas, Babar Shah, Tariq Hussain, Dongho Song, Shaker EI-Sappagh, Jaiteg Singh
Summary: Medical Image Analysis (MIA) is an active research area in computer vision, with brain tumor detection being a major focus due to its severity. However, existing systems may not efficiently classify brain tumors with high accuracy, and there is a lack of smart and easily implementable approaches in 2D and 3D medical images. This paper proposes a novel two-tier framework for tumor detection and localization in MRI, and introduces a well-annotated dataset. Experimental results demonstrate the effectiveness of the proposed framework, achieving 97% accuracy for classification and 83% for localization.
CMC-COMPUTERS MATERIALS & CONTINUA
(2022)
Article
Computer Science, Artificial Intelligence
Cong Liu, Xiaolong Xu
Summary: In this paper, we propose an attention-based multi-feature fusion method (AMFF) for intention recognition in short text, which addresses the data sparsity issue. By enriching short text features through the fusion of TF-IDF, CNNs, and LSTM extracted features, as well as using attention mechanisms to measure important features, the experimental results demonstrate that the AMFF model outperforms traditional and typical deep learning models in short text classification.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Information Systems
Souheil Fenghour, Daqing Chen, Kun Guo, Bo Li, Perry Xiao
Summary: This paper presents a survey on automated lip-reading approaches, focusing on deep learning methodologies. The survey compares different components of automated lip-reading systems and highlights the advantages of Convolutional Neural Networks, Attention-Transformers, and Temporal Convolutional Networks. Additionally, it compares different classification schemas used for lip-reading and reviews the most up-to-date lip-reading systems.
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
Chemistry, Analytical
Chunpeng Gong, Aijuan Li, Yumin Song, Ning Xu, Weikai He
Summary: This paper proposes an improved YOLOv3-based traffic sign recognition method, which solves the problems of small traffic signs, inconspicuous characteristics, and low detection accuracy through the fusion of local and global features, introduction of additional feature prediction scale, and use of DIoU loss. It achieves higher accuracy in traffic sign detection, especially for small targets, while maintaining real-time performance.
Article
Engineering, Electrical & Electronic
Hongbo Yang, Ping Liu, YuZhen Hu, JingNan Fu
Summary: The paper discusses the importance of object recognition and classification in underwater videos, and uses YOLOv3 algorithm and Faster R-CNN algorithm to detect marine objects, with results showing that YOLOv3 algorithm outperforms Faster R-CNN algorithm in both speed and accuracy.
MICROSYSTEM TECHNOLOGIES-MICRO-AND NANOSYSTEMS-INFORMATION STORAGE AND PROCESSING SYSTEMS
(2021)