2024

• Record 409 of
  
  Title:A Cross-Level Interaction Network Based on Scale-Aware Augmentation for Camouflaged Object Detection
  
  Author Full Names:Ma, Ming; Sun, Bangyong

  Source Title:IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE

  Language:English

  Document Type:Article

  Abstract:Camouflaged object detection (COD), with the task of separating the camouflaged object from its color/texture similar background, has been widely used in the fields of medical diagnosis and military reconnaissance. However, the COD task is still a challenging problem due to two main difficulties: large scale-variation for different camouflaged objects, and extreme similarity between the camouflaged object and its background. To address these problems, a cross-level interaction network based on scale-aware augmentation (CINet) for the COD task is proposed. Specifically, a scale-aware augmentation module (SAM) is firstly designed to perceive the scales information of the camouflaged object by calculating an optimal receptive field. Furthermore, a cross-level interaction module (CLIM) is proposed to facilitate the interaction of scale information at all levels, and the context of the feature maps is enriched accordingly. Finally, with the purpose of fully utilizing these features, we design a dual-branch feature decoder (DFD) to strengthen the connection between the predictions at each level. Extensive experiments performed on four CODdatasets, e.g., CHAMELEON, CAMO, COD10K, and NC4K, demonstrate the superiority of the proposed CINet compared with 21 existing state-of-the-art methods.

  Addresses:[Ma, Ming; Sun, Bangyong] Xian Univ Technol, Sch Printing Packaging & Digital Media, Xian 710048, Peoples R China; [Sun, Bangyong] Chinese Acad Sci, Xian Inst Opt & Precis Mech, Key Lab Spectral Imaging Technol CAS, Xian 710119, Peoples R China

  Affiliations:Xi'an University of Technology; Chinese Academy of Sciences; Xi'an Institute of Optics & Precision Mechanics, CAS

  Publication Year:2024

  Volume:8

  Issue:1

  Start Page:69

  End Page:81

  DOI Link:http://dx.doi.org/10.1109/TETCI.2023.3299305

  数据库ID（收录号）:WOS:001051266200001
• Record 410 of
  
  Title:RGB-guided hyperspectral image super-resolution with deep progressive learning
  
  Author Full Names:Zhang, Tao; Fu, Ying; Huang, Liwei; Li, Siyuan; You, Shaodi; Yan, Chenggang

  Source Title:CAAI TRANSACTIONS ON INTELLIGENCE TECHNOLOGY

  Language:English

  Document Type:Article

  Keywords Plus:CLASSIFICATION; RESOLUTION; SYSTEM

  Abstract:Due to hardware limitations, existing hyperspectral (HS) camera often suffer from low spatial/temporal resolution. Recently, it has been prevalent to super-resolve a low resolution (LR) HS image into a high resolution (HR) HS image with a HR RGB (or multispectral) image guidance. Previous approaches for this guided super-resolution task often model the intrinsic characteristic of the desired HR HS image using hand-crafted priors. Recently, researchers pay more attention to deep learning methods with direct supervised or unsupervised learning, which exploit deep prior only from training dataset or testing data. In this article, an efficient convolutional neural network-based method is presented to progressively super-resolve HS image with RGB image guidance. Specifically, a progressive HS image super-resolution network is proposed, which progressively super-resolve the LR HS image with pixel shuffled HR RGB image guidance. Then, the super-resolution network is progressively trained with supervised pre-training and unsupervised adaption, where supervised pre-training learns the general prior on training data and unsupervised adaptation generalises the general prior to specific prior for variant testing scenes. The proposed method can effectively exploit prior from training dataset and testing HS and RGB images with spectral-spatial constraint. It has a good generalisation capability, especially for blind HS image super-resolution. Comprehensive experimental results show that the proposed deep progressive learning method outperforms the existing state-of-the-art methods for HS image super-resolution in non-blind and blind cases.

  Addresses:[Zhang, Tao; Fu, Ying] Beijing Inst Technol, Sch Comp Sci & Technol, Beijing, Peoples R China; [Huang, Liwei] Beijing Inst Remote Sensing, Satellite Informat Intelligent Proc & Applicat Res, Beijing, Peoples R China; [Li, Siyuan] Chinese Acad Sci, Xian Inst Opt & Precis Mech, Key Lab Spectral Imaging Technol, Xian, Peoples R China; [You, Shaodi] Univ Amsterdam, Inst Informat, Amsterdam, Netherlands; [Yan, Chenggang] Hangzhou Dianzi Univ, Sch Commun Engn, Hangzhou, Peoples R China

  Affiliations:Beijing Institute of Technology; Chinese Academy of Sciences; Xi'an Institute of Optics & Precision Mechanics, CAS; University of Amsterdam; Hangzhou Dianzi University

  Publication Year:2024

  Volume:9

  Issue:3

  Start Page:679

  End Page:694

  DOI Link:http://dx.doi.org/10.1049/cit2.12256

  数据库ID（收录号）:WOS:001027404900001
• Record 411 of
  
  Title:Detecting the Background-Similar Objects in Complex Transportation Scenes
  
  Author Full Names:Sun, Bangyong; Ma, Ming; Yuan, Nianzeng; Li, Junhuai; Yu, Tao

  Source Title:IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS

  Language:English

  Document Type:Article

  Keywords Plus:OBSERVABILITY ANALYSIS; CALIBRATION; INTEGRATION; INS; SYSTEMS; ROBUST; GNSS; RTK; BDS

  Abstract:With the development of intelligent transportation systems, most human objects can be accurately detected in normal road scenes. However, the detection accuracy usually decreases sharply when the pedestrians are merged into the background with very similar colors or textures. In this paper, a camouflaged object detection method is proposed to detect the pedestrians or vehicles from the highly similar background. Specifically, we design a guide-learning-based multi-scale detection network (GLNet) to distinguish the weak semantic distinction between the pedestrian and its similar background, and output an accurate segmentation map to the autonomous driving system. The proposed GLNet mainly consists of a backbone network for basic feature extraction, a guide-learning module (GLM) to generate the principal prediction map, and a multi-scale feature enhancement module (MFEM) for prediction map refinement. Based on the guide learning and coarse-to-fine strategy, the final prediction map can be obtained with the proposed GLNet which precisely describes the position and contour information of the pedestrians or vehicles. Extensive experiments on four benchmark datasets, e.g., CHAMELEON, CAMO, COD10K, and NC4K, demonstrate the superiority of the proposed GLNet compared with several existing state-of-the-art methods.

  Addresses:[Sun, Bangyong; Ma, Ming] Xian Univ Technol, Sch Printing Packaging & Digital Media, Xian 710048, Peoples R China; [Yuan, Nianzeng; Li, Junhuai] Xian Univ Technol, Sch Comp Sci & Engn, Xian 710048, Peoples R China; [Yu, Tao] Chinese Acad Sci, Xian Inst Opt & Precis Mech, Key Lab Spectral Imaging Technol CAS, Xian 710119, Peoples R China

  Affiliations:Xi'an University of Technology; Xi'an University of Technology; Chinese Academy of Sciences; Xi'an Institute of Optics & Precision Mechanics, CAS

  Publication Year:2024

  Volume:25

  Issue:3

  Start Page:2920

  End Page:2932

  DOI Link:http://dx.doi.org/10.1109/TITS.2023.3268378

  数据库ID（收录号）:WOS:000980401000001
• Record 412 of
  
  Title:Efficient dense attention fusion network with channel correlation loss for road damage detection
  
  Author Full Names:Liu, Zihan; Jing, Kaifeng; Yang, Kai; Zhang, ZhiJun; Li, Xijie

  Source Title:IET INTELLIGENT TRANSPORT SYSTEMS

  Language:English

  Document Type:Article

  Keywords Plus:PAVEMENT CRACK DETECTION; OBJECT DETECTION; NEURAL-NETWORKS

  Abstract:Road damage detection (RDD) is critical to society's safety and the efficient allocation of resources. Most road damage detection methods which directly adopt various object detection models face some significant challenges due to the characteristics of the RDD task. First, the damaged objects in the road images are highly diverse in scales and difficult to differentiate, making it more challenging than other tasks. Second, existing methods neglect the relationship between the feature distribution and model structure, which makes it difficult for optimization. To address these challenges, this study proposes an efficient dense attention fusion network with channel correlation loss for road damage detection. First, the K-Means++ algorithm is applied for data preprocessing to optimize the initial cluster centers and improve the model detection accuracy. Second, a dense attention fusion module is proposed to learn spatial-spectral attention to enhance multi-scale fusion features and improve the ability of the model to detect damage areas at different scales. Third, the channel correlation loss is adopted in the class prediction process to maintain the separability of intra and inter-class. The experimental results on the collected RDDA dataset and RDD2022 dataset show that the proposed method achieves state-of-the-art performance.

  Addresses:[Liu, Zihan; Jing, Kaifeng] AmazingX Acad, Foshan, Peoples R China; [Yang, Kai; Zhang, ZhiJun; Li, Xijie] Wuhan Univ Technol, Sch Comp Sci & Artificial Intelligence, Wuhan, Peoples R China; [Yang, Kai; Li, Xijie] Wuhan Univ Technol, Sanya Sci & Educ Innovat Pk, Sanya, Peoples R China; [Li, Xijie] Xian Inst Opt & Precis Mech CAS, Xian 710119, Peoples R China

  Affiliations:Wuhan University of Technology; Wuhan University of Technology; Chinese Academy of Sciences; Xi'an Institute of Optics & Precision Mechanics, CAS

  Publication Year:2024

  Volume:18

  Issue:10

  Start Page:1747

  End Page:1759

  DOI Link:http://dx.doi.org/10.1049/itr2.12369

  数据库ID（收录号）:WOS:000972343700001