Jingli Gao

Associate Professor, School of Software Engineering, Pingdingshan University

 

Link

Email

Github

Contact

Room West 505, Science and Technology Building, Pingdingshan University, Pingdingshan City 467000, China

    [2024]

  1. A segmentation network based on residual blocks and multi-channel images, 2024 3rd International Conference on Image Processing and Media Computing, 2024. [BibTeX][PDF][Abstract]
    Jingli Gao, Mengya Zhang,Li Ma,Miao Huang, Zhen Li.
  2. BibTeX:
    @INPROCEEDINGS{gao-etal-2024-seg-weld,
    	author={Gao, Jingli and Zhang, Mengya and Ma, Li and Huang, Miao and Li, Zhen},
    	booktitle={2024 3rd International Conference on Image Processing and Media Computing (ICIPMC)}, 
    	title={A segmentation network based on residual blocks and multi-channel images}, 
    	year={2024},
    	volume={},
    	number={},
    	pages={137-143},
    	keywords={Location awareness;Image segmentation;Sensitivity;Welding;Media;Inspection;Feature extraction;image segmentation;weld defects;quality inspection;UNet},
    	doi={10.1109/ICIPMC62364.2024.10586689}}
    
    Abstract: As an indispensable key process in the industrial chain, the demand of welding could continue to maintain stable growth. However, various defects could inevitably occur in the welding process, and undetected defects could directly affect the bearing capacity and service life of the welding device. Therefore, it is necessary to develop intelligent detection algorithms for non-destructive quality inspection of the welding products. In this paper, we convert the localization problem and category classification problem of weld defects into a pixel classification problem, and propose a u-shaped segmentation network based on residual blocks and the concatenated multi-channel images, called CatResUNet. First, we concatenate the original weld image and the reversed image to enrich the input data, and obtain a doubled-channel image. We then use residual blocks as the backbone of the encoder of the u-shaped network to extract the weld defect features from the obtained doubled-channel image, and use the skip connection to fuse the underlying features from the encoder and the high-level features of the decoder to ensure the accuracy and detail retention ability of defect segmentation. Finally, we use the convolutional layers instead of the fully connected layers to perform pixel-level classification. Experimental results demonstrate that appending the reversed image to the original image to expand the network’s input channels could enrich the information that was fed into the network, hence effectively reducing the rate of missed and false detections, it is a crucial aspect for weld defect detection. Compared with CatUNet which is a standard u-shaped network based on doubled-channel images, the introduction of the residual blocks can help our CatResUNet network to achieve better or basically equivalent evaluation indicators. Furthermore, compared with the other three approaches, the higher dice coefficient and IOU values achieved by our CatResUNet network indicate its enhanced sensitivity to small targets while maintaining excellent performance for large target segmentation tasks. This ensures that while preserving the model’s ability to segment large defects accurately, it also improves its sensitivity to small defects.

    [2023]

  3. A unified binary classification network for weld image detection, 2023 8th International Conference on Control, Robotics and Cybernetics, 2023. [BibTeX][PDF][Abstract]
    Jingli Gao, Mengya Zhang,Li Ma,Miao Huang, Zhen Li.
  4. BibTeX:
    @INPROCEEDINGS{gao-etal-2023-class-weld,
    	author={Gao, Jingli and Zhang, Mengya and Ma, Li and Huang, Miao and Li, Zhen},
    	booktitle={2023 8th International Conference on Control, Robotics and Cybernetics (CRC)}, 
    	title={A unified binary classification network for weld image detection}, 
    	year={2024},
    	volume={},
    	number={},
    	pages={275-279},
    	keywords={Convolution;Shape;Welding;Transfer learning;Manuals;Inspection;Feature extraction;weld image detection;transfer learning;binary classification;quality inspection},
    	doi={10.1109/CRC60659.2023.10488660}}
    
    Abstract: Quality inspection is one of the important measures to ensure the quality of welding products. In the generation process, there is a lack of weld defects, irregular shape of defects, and unbalanced defect categories, which makes the detection of weld images difficult. At present, the quality inspection of welding products is still carried out by manual inspection of their X-ray images. It is urgent to solve the primary problem of detection that determines whether there is a defect in a weld image. In this paper, we propose a weld image detection method which is converted into a binary classification problem. Our method consists of a feature extraction module and a classification module. The feature extraction module uses residual network to extract feature map based on model parameters learned from large-scale ImageNet dataset, which is fine-tuned on weld images. The classification module combines global average pooling and convolution layers instead of fully collected layer to classify weld images. Experiments show that our proposed method achieves 97.65% true positive rate and 99.85% true negative rate on the weld image dataset. and has better generalization performance in comparison with other methods.

    [2022]

  5. A comprehensive social matrix factorization for recommendations with prediction and feedback mechanisms by fusing trust relationships and social tags, Soft Computing, 2022, (8): 1-18. [BibTeX][PDF][Abstract]
    Rui Chen, Jianwei Zhang, Zhifeng Zhang, Yanshuo Chang, Jingli Gao, Pu Li, Hui Liang.
  6. BibTeX:
    @article{wang-etal-2022-social,
      author={Rui Chen, Jianwei Zhang, Zhifeng Zhang, Yanshuo Chang, Jingli Gao, Pu Li, Hui Liang.},
      journal={Soft Computing}, 
      title={A comprehensive social matrix factorization for recommendations with prediction and feedback mechanisms by fusing trust relationships and social tags}, 
      year={2022},
      volume={},
      number={8},
      pages={1-18},
      doi={https://doi.org/10.1007/s00500-022-07440-x}
    }
    
    Abstract: Social relationships play an important role in improving the quality of recommender systems (RSs). A large number of experimental results show that social relationship-based recommendation methods alleviate the problems of data sparseness and cold start in RSs to some extent. However, existing recommendation methods have difficulty in accurately obtaining user features and item features, which seriously affects recommendation system performance. To accurately model social relationships and improve recommendation quality, we use both explicit (e.g. user-item ratings, trust relationships) and implicit (e.g. social tags) social relationships to mine users’ potential interest preferences; thus, we propose a social recommendation method incorporating trust relationships and social tags. The method maps user features and item features to a shared feature space using the above social relationship, obtains user similarity and item similarity through potential feature vectors of users and items, and continuously trains them to obtain accurate similarity relationships to improve recommendation performance. The experimental results demonstrate that our proposed approach achieves superior performance over the other social recommendation approaches.
  7. A Sentiment Classification Method of Web Social Media Based on Multi-dimension and Multi-level Modeling, IEEE Transactions on Industrial Informatics (T-II), 2022. [BibTeX][PDF][Abstract]
    Bingkun Wang, Donghong Shan, Aiwan Fan, Lei Liu, Jingli Gao.
  8. BibTeX:
    @article{wang-etal-2022-sentiment,
      author={Wang, Bingkun and Shan, Donghong and Fan, Aiwan and Liu, Lei and Gao, Jingli},
      journal={IEEE Transactions on Industrial Informatics}, 
      title={A Sentiment Classification Method of Web Social Media Based on Multidimensional and Multilevel Modeling}, 
      year={2022},
      volume={18},
      number={2},
      pages={1240-1249},
      doi={10.1109/TII.2021.3085663}
    }
    
    Abstract: Sentiment classification of web social media faces the problem of text context semantics missing. The existing research mainly solves the problem of text context semantic missing by mining language symbol information in web social media text, seldom considering the emoticon symbols and punctuation symbols in web social media text. Similar to language symbols, emoticons’ symbols and punctuation symbols in web social media text also contain certain sentiment information. In order to make full use of sentiment information contained in web social media to solve the problem of text context semantics missing, we propose a sentiment classification method of web social media based on multidimensional and multilevel modeling. By modeling web social media text from three dimensions (language symbols, emoticons’ symbols, and punctuation symbols) and three levels (words, sentences, and documents) based on a deep learning framework, in this article, we attempt to solve text context semantics missing faced by the sentiment classification of web social media and improve the accuracy of sentiment classification of web social media. The experimental results on Sina Weibo and Twitter datasets show that the average accuracy of our method is 0.9479, which achieves more than 5.86% performance compared with the existing sentiment classification methods.

    [2017]

  9. Robust Small Target Co-Detection from Airborne Infrared Image Sequences, Sensors, 2017. [BibTeX][PDF][Abstract]
    Jingli Gao, Chenglin Wen, Meiqin Liu.
  10. BibTeX:
    @article{gao-etal-2017-co-detection,
    	author = {Gao, Jingli and Wen, Chenglin and Liu, Meiqin},
    	title = {Robust Small Target Co-Detection from Airborne Infrared Image Sequences},
    	journal = {Sensors},
    	year = {2017},
    	volume = {17},
    	number = {10},
    	pages = {2242},
    	doi = {10.3390/s17102242}
    }
    
    Abstract: In this paper, a novel infrared target co-detection model combining the self-correlation features of backgrounds and the commonality features of targets in the spatio-temporal domain is proposed to detect small targets in a sequence of infrared images with complex backgrounds. Firstly, a dense target extraction model based on nonlinear weights is proposed, which can better suppress background of images and enhance small targets than weights of singular values. Secondly, a sparse target extraction model based on entry-wise weighted robust principal component analysis is proposed. The entry-wise weight adaptively incorporates structural prior in terms of local weighted entropy, thus, it can extract real targets accurately and suppress background clutters efficiently. Finally, the commonality of targets in the spatio-temporal domain are used to construct target refinement model for false alarms suppression and target confirmation. Since real targets could appear in both of the dense and sparse reconstruction maps of a single frame, and form trajectories after tracklet association of consecutive frames, the location correlation of the dense and sparse reconstruction maps for a single frame and tracklet association of the location correlation maps for successive frames have strong ability to discriminate between small targets and background clutters. Experimental results demonstrate that the proposed small target co-detection method can not only suppress background clutters effectively, but also detect targets accurately even if with target-like interference.
  11. A two-layer detection model for infrared slow lowaltitude targets, CAC, 2017. [BibTeX][PDF][Abstract]
    Jingli Gao, Chenglin Wen, Meiqin Liu.
  12. BibTeX:
    @inproceedings{gao-etal-2017-two-layer,
      author={Gao, Jingli and Wen, Chenglin and Liu, Meiqin},
      booktitle={2017 Chinese Automation Congress (CAC)}, 
      title={A two-layer detection model for infrared slow low-altitude targets}, 
      year={2017},
      volume={},
      number={},
      pages={7168-7173},
      doi={10.1109/CAC.2017.8244071}
    }
    
    Abstract: This paper proposes a novel detection approach for dim targets with low signal-to-noise ratios in an image sequence. Initially, the superposition analysis is introduced to reveal the relationship between target energy and noise energy in the overlapped images, which is vital for the effectiveness of singular value decomposition, and also the relationship of signal-to-noise ratios to angles between singular value vectors is analyzed, which illustrates the essence of angle-based detection methods. Second, analyzing the feasibility of locating targets using singular vectors, thus the first few singular vectors and threshold technology are combined to reconstruct the targets in each overlapped image, and then the positions of the suspected targets are connected to form tracks, which is validated in terms of Hough transform. Extensive experiments show that the proposed method not only works more stably under different signal-to-noise ratios, but also has better detection performance compared with the conventional baseline methods.

    [2016]

  13. Detecting slowly moving infrared targets using temporal filtering and association strategy, Frontiers of Information Technology & Electronic Engineering, 2016. [BibTeX][PDF][Abstract][More]
    Jingli Gao, Chenglin Wen, Zhejing Bao, Meiqin Liu.
  14. BibTeX:
    @article{gao-etal-2016-detecting,
      author    = {Gao, Jingli and Wen, Chenglin and Bao, Zhejing and Liu, Meiqin Liu},
      title     = {Detecting slowly moving infrared targets using temporal filtering
                   and association strategy},
      journal   = {Frontiers Inf. Technol. Electron. Eng.},
      volume    = {17},
      number    = {11},
      pages     = {1176--1185},
      year      = {2016},
      doi       = {10.1631/FITEE.1601203},
    }
    
    Abstract: The special characteristics of slowly moving infrared targets, such as containing only a few pixels, shapeless edge, low signal-to-clutter ratio, and low speed, make their detection rather difficult, especially when immersed in complex backgrounds. To cope with this problem, we propose an effective infrared target detection algorithm based on temporal target detection and association strategy. First, a temporal target detection model is developed to segment the interested targets. This model contains mainly three stages, i.e., temporal filtering, temporal target fusion, and cross-product filtering. Then a graph matching model is presented to associate the targets obtained at different times. The association relies on the motion characteristics and appearance of targets, and the association operation is performed many times to form continuous trajectories which can be used to help disambiguate targets from false alarms caused by random noise or clutter. Experimental results show that the proposed method can detect slowly moving infrared targets in complex backgrounds accurately and robustly, and has superior detection performance in comparison with several recent methods.

    [2015]

  15. 基于奇异值分解和叠加法的慢速小目标检测算法, 上海交通大学学报, 2015. [BibTeX][PDF][Abstract]
    高敬礼, 文成林, 刘妹琴.
  16. BibTeX:
    @article{gao-etal-2015-svd,
    	author={高敬礼 and 文成林 and 刘妹琴},
    	title={ 基于奇异值分解和叠加法的慢速小目标检测算法 },
    	journal={上海交通大学学报},
    	year={2015},
    	volume={49},
    	number={6},
    	pages={876-883},
    }
    

    [2014]

  17. Steel surface defect detection and localization based on SVD and two-side compressive measurements, CCDC, 2014. [BibTeX][PDF][Abstract]
    Jingli Gao, Chenglin Wen, Meiqin Liu.
  18. BibTeX:
    @inproceedings{gao-etal-2014-steel,
      author={Gao, Jingli and Wen, Chenglin and Liu, Meiqin},
      booktitle={The 26th Chinese Control and Decision Conference (2014 CCDC)}, 
      title={Steel surface defect detection and localization based on SVD and two-side compressive measurements}, 
      year={2014},
      volume={},
      number={},
      pages={1401-1406},
      doi={10.1109/CCDC.2014.6852386}
    }
    
    Abstract: This paper proposes a method for defect detection and localization based on singular value decomposition and two-side compressive measurements. First, the feasibility of the singular value decomposition for defect detection and localization is analyzed, then the invariance of the geometrical structure of the rows or columns of the raw data and the compressive data is justified, so the energy and pattern contained in the raw data can be transferred into the compressive data and kept in the singular values and singular vectors. On this basis, the proposed defect detection algorithm based on the singular values of compressive data and the proposed defect localization algorithm based on the singular vectors are given without reconstruction of images. Simulation results show that the proposed method based on compressive measurements has a good performance.
  19. SVD-based scattered small targets detection, MFI, 2014. [BibTeX][PDF][Abstract]
    Jingli Gao, Chenglin Wen, Meiqin Liu.
  20. BibTeX:
    @inproceedings{gao-etal-2014-svd,
      author={Gao, Jingli and Wen, Chenglin and Liu, Meiqin},
      booktitle={2014 International Conference on Multisensor Fusion and Information Integration for Intelligent Systems (MFI)}, 
      title={SVD-based scattered small targets detection}, 
      year={2014},
      volume={},
      number={},
      pages={1-6},
      doi={10.1109/MFI.2014.6997759}
    }
    
    Abstract: Due to lack of sufficient shape and strength information about the targets present in the entire image obtained in a wide range of scenarios over a long distance, it is very hard to detect these targets using the existing detection methods. In order to solve this problem, this paper proposes a scattered small target detection algorithm based on singular value decomposition. First, the detectability of a single target is analyzed from the perspective of singular values; second, the detectability of scattered targets is analyzed in terms of singular values and elementary transformations, based on the composability of dispersed targets, the dispersed targets can be combined into a large target using elementary transformations, on this basis, the conclusion is given that the image containing dispersed targets and the image containing a large target have the same detectability; then perform singular value decomposition on the image that may contain dispersed targets and the standard residual image which does not contain targets, and obtain their respective singular value vectors, use these singular value vectors to calculate the cosine angle which was used to determine whether the image contains targets or not, if it is confirmed, the targets are roughly located based on the singular vectors. Finally, the effectiveness of the algorithm is verified using Monte Carlo simulation.