Article
Computer Science, Artificial Intelligence
Tingting Dan, Xijie Chen, Miao He, Hongmei Guo, Xiaoqin He, Jiazhou Chen, Jianbo Xian, Yu Hu, Bin Zhang, Nan Wang, Hongning Xie, Hongmin Cai
Summary: Researchers have developed an automatic DeepGA model for accurate prediction of gestational age. The model uses deep segmentation to identify and segment critical tissues, and then employs deep regression to estimate gestational age. Experimental results demonstrate that DeepGA outperforms traditional measurement methods.
ARTIFICIAL INTELLIGENCE IN MEDICINE
(2023)
Article
Chemistry, Analytical
Xiaowei He, Rao Cheng, Zhonglong Zheng, Zeji Wang
Summary: An improved algorithm named YOLO-MXANet is proposed to enhance the detection accuracy of small objects in traffic scenes. By utilizing CIoU, a lightweight backbone network called SA-MobileNeXt, and other methods, the algorithm is able to improve object detection accuracy and reduce model complexity effectively.
Article
Computer Science, Information Systems
Zorana Dozdor, Tomislav Hrkac, Karla Brkic, Zoran Kalafatic
Summary: This paper conducts a comparative study on the current techniques for automated age estimation from face images, focusing on lightweight models suitable for embedded implementation. The study investigates modern deep learning architectures for feature extraction and different ways of framing the problem as classification, regression, or soft label classification. The models are evaluated on the Audience dataset for age group classification and the FG-NET dataset for exact age estimation. The paper proposes a novel loss function that combines regression and classification approaches and demonstrates its superior performance compared to other methods. Moreover, the lightweight architecture is suitable for implementation on embedded devices.
Article
Computer Science, Interdisciplinary Applications
Zijian Wang, Zixiang Cai, Yimin Wu
Summary: Tunnel construction sites pose safety risks due to low-light conditions. This study proposes an improved YOLOX approach and a new dataset for detecting low-light and small personal protective equipment (PPE). The improved YOLOX achieves higher accuracy and real-time processing speed, reducing safety incidents on tunnel construction sites.
JOURNAL OF COMPUTATIONAL DESIGN AND ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Yang Liu, Peng Sun, Nickolas Wergeles, Yi Shang
Summary: This paper reviews deep learning methods for small object detection, discussing challenges, solutions, and techniques. Experimental results show that Faster R-CNN performs the best in detecting small objects.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Jun Zhang, Yizhen Meng, Zhipeng Chen
Summary: This paper proposes a small target detection method based on deep learning, which improves the current deep network architecture and introduces multiple features and multi-scale detection to effectively solve the problem of small target detection and expand the training dataset effectively, solving the difficulties in labeling small target data and overfitting.
Article
Computer Science, Artificial Intelligence
Sihan Ma, Jizhizi Li, Jing Zhang, He Zhang, Dacheng Tao
Summary: P3M-10k is the first large-scale anonymized benchmark for Privacy-Preserving Portrait Matting, consisting of face-blurred portrait images and alpha mattes. P3M-Net is a unified matting model compatible with both CNN and transformer backbones. The Copy and Paste strategy improves the cross-domain generalization ability.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2023)
Article
Computer Science, Artificial Intelligence
Gyoung S. Na
Summary: This paper proposes a new hierarchical approach to learning rate adaptation in gradient methods, called learning rate optimization (LRO). LRO optimizes the learning rate based on the alternating direction method of multipliers (ADMM), without requiring second-order information and probabilistic models, and without any additional hyperparameters. In experiments, SGD and Adam with LRO outperformed other methods in image classification tasks.
Article
Agriculture, Multidisciplinary
Yunong Tian, Shihui Wang, En Li, Guodong Yang, Zize Liang, Min Tan
Summary: This paper proposes a new method called Multi-scale Dense YOLO (MD-YOLO) for detecting small target lepidopteran pests on sticky insect boards. Experimental results demonstrate the superiority of MD-YOLO in detection accuracy and its practical applicability in real-world field scenes.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Chemistry, Analytical
Zeji Wang, Xiaowei He, Yi Li, Qinliang Chuai
Summary: This article introduces a new convolutional transformer model EmbedFormer, which enhances model performance by introducing DwConv as the token mixer, and achieves excellent results in various tasks.
Article
Engineering, Chemical
Xue Wu, Yanmei Meng, Jinlai Zhang, Jing Wei, Xulei Zhai
Summary: The growth of cane sugar crystals is crucial for the production process. However, the overlap of sugar particles hampers the accurate identification of crystal shapes from images. To address this, we propose ASSugarNet, a deep neural network-based amodal segmentation method for cane sugar crystals. The network incorporates a local feature enhance module and a foreground boundary enhancement loss function to improve segmentation quality. It also uses an amodal segmentation module to predict contours and masks for occluded crystals. Experimental results demonstrate that ASSugarNet outperforms other networks in crystal segmentation, achieving an MIoU of 81.47% and an OA of 91.00%. The AP for adhesive particle segmentation is 32.25%, with an AP50 of 52.14%.
JOURNAL OF FOOD ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Roman Solovyev, Weimin Wang, Tatiana Gabruseva
Summary: This study introduces a novel method, weighted boxes fusion, for combining predictions from different object detection models, significantly improving the quality of the ensemble predicted rectangles. The method achieved top results in various datasets and challenges, with the 3D version of boxes fusion being successfully applied in winning teams of specific competitions.
IMAGE AND VISION COMPUTING
(2021)
Article
Mathematical & Computational Biology
Qihan Feng, Xinzheng Xu, Zhixiao Wang
Summary: Small object detection is important for various real-world applications, and it remains a challenging task in computer vision due to its low resolution and noise representation. This paper focuses on the difficulties of small object detection and analyzes deep learning-based research papers from multiple perspectives. The authors also review literature on crucial tasks such as small face detection and aerial image object detection. Experimental results show that network configuration to enhance the resolution of input features can significantly improve performance. Several potential future research directions in the field of small object detection are also provided.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2023)
Article
Computer Science, Information Systems
Luis Gonzalez-Naharro, M. Julia Flores, Jesus Martinez-Gomez, Jose M. Puerta
Summary: Aesthetic assessment evaluates image quality using subjective annotations. This study presents a new approach based on redefining the rating-based groundtruths to reduce uncertainty and automatically group images into high and low quality patterns. Experimental results on the AVA dataset show significant performance gains, achieving 3% to 9% more balanced accuracy than baseline groundtruths.
INFORMATION PROCESSING & MANAGEMENT
(2023)
Article
Computer Science, Artificial Intelligence
Shuai Luo, Yujie Li, Pengxiang Gao, Yichuan Wang, Seiichi Serikawa
Summary: This paper reviews state-of-the-art image segmentation methods based on meta-learning, introducing the background and differences with other similar methods, discussing various types of meta-learning methods and their applications in image segmentation, conducting experimental comparisons, and highlighting future trends of meta-learning in image segmentation.
PATTERN RECOGNITION
(2022)
Article
Pediatrics
Laura Sand, Lisa Szatkowski, T'ng Chang Kwok, Don Sharkey, David A. Todd, Helen Budge, Shalini Ojha
Summary: The analysis of the National Neonatal Research Database from 2010-17 revealed a significant increase in the use of NIV, particularly high-flow nasal cannula, which was associated with reduced odds of death before discharge but increased odds of bronchopulmonary dysplasia (BPD) and other adverse outcomes. The study emphasizes the need for robust clinical evidence to improve outcomes with the use of NIV as initial and ongoing respiratory support, as NIV use is increasing, especially as initial respiratory support.
ARCHIVES OF DISEASE IN CHILDHOOD-FETAL AND NEONATAL EDITION
(2022)
Article
Pediatrics
Lara Shipley, Gillian Hyliger, Don Sharkey
Summary: Between 2011 and 2016, the rate of intra-uterine transfer of extremely preterm infants within the UK decreased, with an associated increase in early postnatal transfer, including early postnatal transfer between level 3 NICUs.
ARCHIVES OF DISEASE IN CHILDHOOD-FETAL AND NEONATAL EDITION
(2022)
Article
Pediatrics
Lara Shipley, Aarti Mistry, Don Sharkey
Summary: This study in the UK found that almost half of infants with HIE were born in non-cooling centres, which led to suboptimal hypothermic treatment and reduced seizure-free survival. There is a need to consider providing active TH in non-CC hospitals prior to upward transfer.
ARCHIVES OF DISEASE IN CHILDHOOD-FETAL AND NEONATAL EDITION
(2022)
Article
Physiology
Rebecca J. Calthorpe, Caroline Poulter, Alan R. Smyth, Don Sharkey, Jayesh Bhatt, Gisli Jenkins, Amanda L. Tatler
Summary: As survival rates for extremely preterm infants improve, there is an increase in bronchopulmonary dysplasia (BPD), one of the most significant complications of preterm birth. BPD development is multifactorial, resulting from exposure to various antenatal and postnatal stressors. BPD has short-term and long-term implications, including respiratory, cardiovascular, and neurological issues. The review focuses on the role of transforming growth factor I3 (TGF-I3) in lung development, the impact of known risk factors on the TGF-I3 signaling pathway in BPD, and the effect of current and potential medications on TGF-I3 signaling for BPD prevention and treatment.
AMERICAN JOURNAL OF PHYSIOLOGY-LUNG CELLULAR AND MOLECULAR PHYSIOLOGY
(2023)
Review
Pediatrics
T'ng Chang Kwok, Natalie Batey, Ka Ling Luu, Andrew Prayle, Don Sharkey
Summary: Prediction models can identify infants at high risk of BPD and guide targeted preventative strategies. A systematic review and meta-analysis evaluated available models and validated their performance. Most prediction models are outdated, single-center studies with a high risk of bias.
PEDIATRIC RESEARCH
(2023)
Article
Pediatrics
Lisa Szatkowski, Sheeza Fateh, Janine Abramson, T'ng Chang Kwok, Don Sharkey, Helen Budge, Shalini Ojha
Summary: This study quantifies the trends in caffeine use in preterm infants and investigates the effects of early vs late caffeine on neonatal outcomes. The results show that early caffeine administration is associated with reduced risks of respiratory diseases and brain injury.
ARCHIVES OF DISEASE IN CHILDHOOD-FETAL AND NEONATAL EDITION
(2023)
Article
Chemistry, Analytical
Alexander Turner, Stephen Hayes, Don Sharkey
Summary: Neurodevelopmental delay following extremely preterm birth or birth asphyxia is common but diagnosis is often delayed. Early interventions have been shown to improve outcomes. This study proposes a new method using deep learning and 2D pose estimation to assess children's movements when interacting with toys, which could assist in accurate diagnosis and treatment monitoring.
Article
Respiratory System
T'ng Chang Kwok, Lisa Szatkowski, Don Sharkey
Summary: This retrospective study examined the association between the timing of postnatal dexamethasone (PND) treatment and mortality and respiratory outcomes in preterm infants. The results showed that commencing PND treatment after 5 weeks of age was associated with worse respiratory outcomes. Further clinical trials are needed to determine the optimal timing of PND treatment.
EUROPEAN RESPIRATORY JOURNAL
(2023)
Article
Respiratory System
Tng Chang Kwok, Caroline Poulter, Saleh Algarni, Lisa Szatkowski, Don Sharkey
Summary: This study examines the changing respiratory management and outcomes of preterm infants in England and Wales from 2010 to 2020. It finds an increase in antenatal corticosteroid use and a decrease in neonatal surfactant use. The study also reveals a decrease in mortality but an increase in bronchopulmonary dysplasia and the need for postdischarge respiratory support, indicating the potential development of chronic respiratory diseases in these infants.
Article
Computer Science, Artificial Intelligence
Mani Kumar Tellamekala, Timo Giesbrecht, Michel Valstar
Summary: This paper explores the problem of recognizing apparent emotion from audio-visual signals in naturalistic conditions. By introducing the Affective Processes model and extending it to the speech domain and audio-visual affect recognition, superior performance has been achieved.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Article
Medicine, General & Internal
Lisa Szatkowski, Don Sharkey, Helen Budge, Shalini Ojha
Summary: This study aimed to investigate the impact of opioids use during mechanical ventilation on infants born at <32 weeks' gestational age. The findings suggest that the use of opioids is associated with an increased risk of preterm brain injury in preterm infants.
Article
Computer Science, Artificial Intelligence
Ioanna Ntinou, Enrique Sanchez, Adrian Bulat, Michel Valstar, Georgios Tzimiropoulos
Summary: In this study, a novel approach based on Heatmap Regression is proposed to jointly estimate the localization and intensity of Action Units (AUs). The method utilizes variable size heatmaps to model AUs intensity and employs transfer learning from a network trained on facial landmark datasets. Experimental results demonstrate that the system achieves state-of-the-art performance on three popular datasets, BP4D, DISFA, and FERA2017.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Siyang Song, Shashank Jaiswal, Enrique Sanchez, Georgios Tzimiropoulos, Linlin Shen, Michel Valstar
Summary: This article addresses two important issues in automatic personality analysis systems: the use of short video segments or single frames for inferring personality traits, and the lack of methods for encoding person-specific facial dynamics. To tackle these issues, the paper proposes a novel Rank Loss for self-supervised learning of facial dynamics and a method to represent person-specific dynamics. The approach achieves promising results in personality estimation and shows the importance of the tasks performed by the subject in the video and the use of multi-scale dynamics.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Article
Pediatrics
Chris Gale, Don Sharkey, Kathryn E. Fitzpatrick, Helen Mactier, Alessandra Morelli, Mariko Nakahara, Madeleine Hurd, Anna Placzek, Marian Knight, Shamez N. Ladhani, Elizabeth S. Draper, Cora Doherty, Maria A. Quigley, Jennifer J. Kurinczuk
Summary: This study used national population-level data to investigate the impact of different SARS-CoV-2 variants on neonatal outcomes. It found that neonatal infection with wildtype SARS-CoV-2 is rare and good outcomes are predominant. However, severe neonatal illness was more common during the delta variant period, potentially due to more severe maternal disease and associated preterm birth.
ARCHIVES OF DISEASE IN CHILDHOOD-FETAL AND NEONATAL EDITION
(2023)
Article
Computer Science, Artificial Intelligence
Mani Kumar Tellamekala, Timo Giesbrecht, Michel Valstar
Summary: This study focuses on the uncertainty issue in facial emotion recognition and applies uncertainty to personality recognition tasks. Experimental results demonstrate that the fusion of uncertainty greatly improves the performance of personality recognition.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2022)