Article
Chemistry, Multidisciplinary
Fei Yan, Cheng Chen, Peng Xiao, Siyu Qi, Zhiliang Wang, Ruoxiu Xiao
Summary: This study summarizes the achievements in the field of saliency prediction, including the early neurological and psychological mechanisms, the guiding role of classic models, and the development process and data comparison of classic and deep saliency prediction models. It also discusses the relationship between the model and human vision, the factors causing semantic gaps, the influences of attention in cognitive research, the limitations of the saliency model, and the emerging applications.
APPLIED SCIENCES-BASEL
(2022)
Article
Engineering, Multidisciplinary
Haoran Huang, Gaopeng Zhao, Yuming Bo, Junyan Yu, LiDong Liang, Yi Yang, Kai Ou
Summary: A railway intrusion detection method is proposed for UAV surveillance, which can accurately detect railway intrusion events without pre-setting the intruding object type. It detects the normal railway region and the abnormal railway region with intruding objects instead of recognizing the intruding objects directly.
Article
Automation & Control Systems
Zhongxu Hu, Chen Lv, Peng Hang, Chao Huang, Yang Xing
Summary: The study proposed a method for driver attention estimation based on a dual-view scene and calibration-free gaze direction, which achieved more accurate results using a multiresolution neural network to handle calibration-free features. Experimental results demonstrated that the proposed method outperforms existing technologies and is applicable to different landscapes, times, and weather conditions.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Deyi Wang, Chengkun Zhang, Min Han
Summary: This research introduces MLFC-Net, a multi-level semantic feature clustering attention model based on deep convolution neural networks (DCNNs), which efficiently extracts accurate feature information for remote sensing image scene classification. The model improves the representation of critical semantic aspects and achieves state-of-the-art results on multiple RSSC datasets.
COMPUTERS & GEOSCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Alexandre Bruckert, Hamed R. Tavakoli, Zhi Liu, Marc Christie, Olivier Le Meur
Summary: Deep learning techniques are widely used for modeling human visual saliency, with the choice of the loss function being a key factor that can significantly impact results. This study demonstrates that modifying the loss function on a fixed network architecture can lead to improvements or depreciation in performance. Combining several well-chosen loss functions in a linear combination can lead to significant improvements in performance on different datasets and network architectures.
Article
Computer Science, Information Systems
Phutphalla Kong, Matei Mancas, Bernard Gosselin, Kimtho Po
Summary: The article introduces a new visual attention model called DeepRare2021 (DR21), which combines the advantages of deep learning and feature engineering. Compared to traditional DNN models, DR21 demonstrates higher efficiency and generality in extracting surprising or unusual data. The model does not require additional training and performs well on multiple eye-tracking datasets.
Article
Engineering, Electrical & Electronic
Osama A. Shawky, Ahmed Hagag, El-Sayed A. El-Dahshan, Manal A. Ismail
Summary: This paper proposes a model for VHR scene classification that utilizes convolutional neural networks and saliency detection algorithm to extract features, performs feature fusion and uses an enhanced multilayer perceptron for image classification. Experiments on multiple datasets demonstrate that the proposed model outperforms existing scene classification models.
SIGNAL IMAGE AND VIDEO PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Zheng Xiao, Zhenyu Nie, Chao Song, Anthony Theodore Chronopoulos
Summary: This paper proposes an extended attention-based framework for scene text recognition tasks. By introducing the Attention on Attention (AoA) mechanism, the relevance between attention results and queries can be determined, improving the accuracy of recognition. Experimental results show that the proposed method outperforms other benchmarks on multiple datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Chemistry, Analytical
Rongbin Yi, Jinlong Li, Lin Luo, Yu Zhang, Xiaorong Gao, Jianqiang Guo
Summary: The paper proposes a deep learning-based point cloud registration method, DOPNet, which extracts global features of point clouds with dynamic graph convolutional neural network and cascading offset-attention modules, achieving more accurate and efficient point cloud registration.
Article
Computer Science, Artificial Intelligence
Wenguan Wang, Jianbing Shen, Jianwen Xie, Ming-Ming Cheng, Haibin Ling, Ali Borji
Summary: This research focuses on predicting visual attention in dynamic scenes, introducing a new benchmark DHF1K and a novel video saliency model ACLNet. Through extensive evaluation on multiple datasets and analysis of saliency models, ACLNet shows superior performance and fast processing speed.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Multidisciplinary Sciences
Tomasz Hachaj, Anna Stolinska, Magdalena Andrzejewska, Piotr Czerski
Summary: This study introduces a machine learning method capable of accurately predicting students' visual attention when solving quizzes, achieving better results than current methods. The predictions are moderately positively correlated with actual data, with a coefficient of 0.547 +/- 0.109. Visual analyses of the obtained saliency maps align with the researchers' experience and expectations in the field.
Article
Computer Science, Information Systems
Lamine Benrais, Nadia Baha
Summary: This paper introduces a novel and simple approach of high-level scene classification utilizing objects to construct background knowledge. The most salient objects are identified and used in computing the appropriate scene category. Experimental results are reported and discussed to prove the efficiency of the proposed method.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Fei Yan, Zhiliang Wang, Siyu Qi, Ruoxiu Xiao
Summary: This study proposes a multilevel saliency prediction network that uses a combination of spatial and channel information to find possible high-level features, further improving the performance of a saliency model.
Article
Multidisciplinary Sciences
Kazuaki Akamatsu, Tomohiro Nishino, Yoichi Miyawaki
Summary: This study used a deep convolutional neural network model to extract hierarchical visual features from natural scene images and evaluated the extent to which the human gaze is attracted to these visual features in space and time. The results showed that the human gaze is more strongly attracted to spatial locations containing higher-order visual features, rather than lower-order visual features or locations predicted by conventional saliency. The time course analysis of gaze attraction revealed a prominent bias towards higher-order visual features shortly after the beginning of observation.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Artificial Intelligence
Hamid Rezatofighi, Tianyu Zhu, Roman Kaskman, Farbod T. Motlagh, Javen Qinfeng Shi, Anton Milan, Daniel Cremers, Laura Leal-Taixe, Ian Reid
Summary: This paper addresses the task of set prediction using deep feed-forward neural networks. It presents a novel approach for learning to predict sets with unknown permutation and cardinality using deep neural networks. The validity of the proposed approach is demonstrated on various vision problems.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)