Adversarial attacks pose a significant threat to artificial intelligence systems by exposing them to vulnerabilities in deep learning models.Existing defense mechanisms often suffer drawbacks,such as the need for mode...Adversarial attacks pose a significant threat to artificial intelligence systems by exposing them to vulnerabilities in deep learning models.Existing defense mechanisms often suffer drawbacks,such as the need for model retraining,significant inference time overhead,and limited effectiveness against specific attack types.Achieving perfect defense against adversarial attacks remains elusive,emphasizing the importance of mitigation strategies.In this study,we propose a defense mechanism that applies random cropping and Gaussian filtering to input images to mitigate the impact of adversarial attacks.First,the image was randomly cropped to vary its dimensions and then placed at the center of a fixed 299299 space,with the remaining areas filled with zero padding.Subsequently,Gaussian×filtering with a 77 kernel and a standard deviation of two was applied using a convolution operation.Finally,the×smoothed image was fed into the classification model.The proposed defense method consistently appeared in the upperright region across all attack scenarios,demonstrating its ability to preserve classification performance on clean images while significantly mitigating adversarial attacks.This visualization confirms that the proposed method is effective and reliable for defending against adversarial perturbations.Moreover,the proposed method incurs minimal computational overhead,making it suitable for real-time applications.Furthermore,owing to its model-agnostic nature,the proposed method can be easily incorporated into various neural network architectures,serving as a fundamental module for adversarial defense strategies.展开更多
Scene graph prediction has emerged as a critical task in computer vision,focusing on transforming complex visual scenes into structured representations by identifying objects,their attributes,and the relationships amo...Scene graph prediction has emerged as a critical task in computer vision,focusing on transforming complex visual scenes into structured representations by identifying objects,their attributes,and the relationships among them.Extending this to 3D semantic scene graph(3DSSG)prediction introduces an additional layer of complexity because it requires the processing of point-cloud data to accurately capture the spatial and volumetric characteristics of a scene.A significant challenge in 3DSSG is the long-tailed distribution of object and relationship labels,causing certain classes to be severely underrepresented and suboptimal performance in these rare categories.To address this,we proposed a fusion prototypical network(FPN),which combines the strengths of conventional neural networks for 3DSSG with a Prototypical Network.The former are known for their ability to handle complex scene graph predictions while the latter excels in few-shot learning scenarios.By leveraging this fusion,our approach enhances the overall prediction accuracy and substantially improves the handling of underrepresented labels.Through extensive experiments using the 3DSSG dataset,we demonstrated that the FPN achieves state-of-the-art performance in 3D scene graph prediction as a single model and effectively mitigates the impact of the long-tailed distribution,providing a more balanced and comprehensive understanding of complex 3D environments.展开更多
Recent research on adversarial attacks has primarily focused on white-box attack techniques,with limited exploration of black-box attack methods.Furthermore,in many black-box research scenarios,it is assumed that the ...Recent research on adversarial attacks has primarily focused on white-box attack techniques,with limited exploration of black-box attack methods.Furthermore,in many black-box research scenarios,it is assumed that the output label and probability distribution can be observed without imposing any constraints on the number of attack attempts.Unfortunately,this disregard for the real-world practicality of attacks,particularly their potential for human detectability,has left a gap in the research landscape.Considering these limitations,our study focuses on using a similar color attack method,assuming access only to the output label,limiting the number of attack attempts to 100,and subjecting the attacks to human perceptibility testing.Through this approach,we demonstrated the effectiveness of black box attack techniques in deceiving models and achieved a success rate of 82.68%in deceiving humans.This study emphasizes the significance of research that addresses the challenge of deceiving both humans and models,highlighting the importance of real-world applicability.展开更多
Weizmannia coagulans(formerly Bacillus coagulans)is a spore-forming and lactic acid-producing bacterium.It has recently attracted much attention from researchers and food manufacturers due to its probiotic functions a...Weizmannia coagulans(formerly Bacillus coagulans)is a spore-forming and lactic acid-producing bacterium.It has recently attracted much attention from researchers and food manufacturers due to its probiotic functions and stability in processing and storage.W.coagulans is capable of improving gut health through the regulation of gut microbiota,modulation of immunity,and improving digestibility and metabolism.Spores,germinated cells and metabolites of W.coagulans modulate the gut micro-environment and further affect other organs.W.coagulans is an environment-friendly probiotic since it can contribute to the host by reconstructing the balance of gut microbiota and only temporarily resides in the intestine after administration.W.coagulans has been generally recognized as safe(GRAS)by the US Food and Drug Administration(FDA),thus it is an ideal probiotic for improving gut health.The merit of its stability in processing and storage provides W.coagulans spores many possibilities for its use in various types of functional foods.This review presents an overview of the characteristics of W.coagulans that make it an ideal probiotic candidate and highlights the proposed health benefits with scientific evidence conferred by the administration of W.coagulans.展开更多
Reconstructing a three-dimensional(3D)environment is an indispensable technique to make augmented reality and augmented virtuality feasible.A Kinect device is an efficient tool for reconstructing 3D environments,and u...Reconstructing a three-dimensional(3D)environment is an indispensable technique to make augmented reality and augmented virtuality feasible.A Kinect device is an efficient tool for reconstructing 3D environments,and using multiple Kinect devices enables the enhancement of reconstruction density and expansion of virtual spaces.To employ multiple devices simultaneously,Kinect devices need to be calibrated with respect to each other.There are several schemes available that calibrate 3D images generated frommultiple Kinect devices,including themarker detection method.In this study,we introduce a markerless calibration technique for Azure Kinect devices that avoids the drawbacks of marker detection,which directly affects calibration accuracy;it offers superior userfriendliness,efficiency,and accuracy.Further,we applied a joint tracking algorithm to approximate the calibration.Traditional methods require the information of multiple joints for calibration;however,Azure Kinect,the latest version of Kinect,requires the information of only one joint.The obtained result was further refined using the iterative closest point algorithm.We conducted several experimental tests that confirmed the enhanced efficiency and accuracy of the proposed method for multiple Kinect devices when compared to the conventional markerbased calibration.展开更多
Recently,simultaneous localization and mapping(SLAM)has received considerable attention in augmented reality(AR)libraries and applications.Although the assumption of scene rigidity is common in most visual SLAMs,this ...Recently,simultaneous localization and mapping(SLAM)has received considerable attention in augmented reality(AR)libraries and applications.Although the assumption of scene rigidity is common in most visual SLAMs,this assumption limits the possibilities of AR applications in various real-world environments.In this paper,we propose a new tracking system that integrates SLAMwith amarker detection module for real-time AR applications in static and dynamic environments.Because the proposed system assumes that the marker is movable,SLAM performs tracking andmappingof the static scene except for themarker,andthemarker detector estimates the 3-dimensional pose of the marker attached to the dynamic object.We can place SLAM maps,cameras,and dynamic objects(markers)on the same coordinate system,and we can also accurately augment AR resources in real-time on both the static scene and dynamic objects simultaneously.Additionally,having a static map of the scene in advance has the advantage of being able to performtracking of static and dynamic environments with mapping disabled.In this paper,we evaluate whether the proposed tracking systemis suitable for AR applications and describe a strategy for creating AR applications in static and dynamic environments with a demonstration.展开更多
基金supported by the Glocal University 30 Project Fund of Gyeongsang National University in 2025.
文摘Adversarial attacks pose a significant threat to artificial intelligence systems by exposing them to vulnerabilities in deep learning models.Existing defense mechanisms often suffer drawbacks,such as the need for model retraining,significant inference time overhead,and limited effectiveness against specific attack types.Achieving perfect defense against adversarial attacks remains elusive,emphasizing the importance of mitigation strategies.In this study,we propose a defense mechanism that applies random cropping and Gaussian filtering to input images to mitigate the impact of adversarial attacks.First,the image was randomly cropped to vary its dimensions and then placed at the center of a fixed 299299 space,with the remaining areas filled with zero padding.Subsequently,Gaussian×filtering with a 77 kernel and a standard deviation of two was applied using a convolution operation.Finally,the×smoothed image was fed into the classification model.The proposed defense method consistently appeared in the upperright region across all attack scenarios,demonstrating its ability to preserve classification performance on clean images while significantly mitigating adversarial attacks.This visualization confirms that the proposed method is effective and reliable for defending against adversarial perturbations.Moreover,the proposed method incurs minimal computational overhead,making it suitable for real-time applications.Furthermore,owing to its model-agnostic nature,the proposed method can be easily incorporated into various neural network architectures,serving as a fundamental module for adversarial defense strategies.
基金supported by the Glocal University 30 Project Fund of Gyeongsang National University in 2025.
文摘Scene graph prediction has emerged as a critical task in computer vision,focusing on transforming complex visual scenes into structured representations by identifying objects,their attributes,and the relationships among them.Extending this to 3D semantic scene graph(3DSSG)prediction introduces an additional layer of complexity because it requires the processing of point-cloud data to accurately capture the spatial and volumetric characteristics of a scene.A significant challenge in 3DSSG is the long-tailed distribution of object and relationship labels,causing certain classes to be severely underrepresented and suboptimal performance in these rare categories.To address this,we proposed a fusion prototypical network(FPN),which combines the strengths of conventional neural networks for 3DSSG with a Prototypical Network.The former are known for their ability to handle complex scene graph predictions while the latter excels in few-shot learning scenarios.By leveraging this fusion,our approach enhances the overall prediction accuracy and substantially improves the handling of underrepresented labels.Through extensive experiments using the 3DSSG dataset,we demonstrated that the FPN achieves state-of-the-art performance in 3D scene graph prediction as a single model and effectively mitigates the impact of the long-tailed distribution,providing a more balanced and comprehensive understanding of complex 3D environments.
基金supported by the Research Resurgence under the Glocal University 30 Project at Gyeongsang National University in 2024.
文摘Recent research on adversarial attacks has primarily focused on white-box attack techniques,with limited exploration of black-box attack methods.Furthermore,in many black-box research scenarios,it is assumed that the output label and probability distribution can be observed without imposing any constraints on the number of attack attempts.Unfortunately,this disregard for the real-world practicality of attacks,particularly their potential for human detectability,has left a gap in the research landscape.Considering these limitations,our study focuses on using a similar color attack method,assuming access only to the output label,limiting the number of attack attempts to 100,and subjecting the attacks to human perceptibility testing.Through this approach,we demonstrated the effectiveness of black box attack techniques in deceiving models and achieved a success rate of 82.68%in deceiving humans.This study emphasizes the significance of research that addresses the challenge of deceiving both humans and models,highlighting the importance of real-world applicability.
基金supported by National Natural Science Foundation of China(32172172,32201994)the Foreign Expert Collaboration Project(G2021108010L).
文摘Weizmannia coagulans(formerly Bacillus coagulans)is a spore-forming and lactic acid-producing bacterium.It has recently attracted much attention from researchers and food manufacturers due to its probiotic functions and stability in processing and storage.W.coagulans is capable of improving gut health through the regulation of gut microbiota,modulation of immunity,and improving digestibility and metabolism.Spores,germinated cells and metabolites of W.coagulans modulate the gut micro-environment and further affect other organs.W.coagulans is an environment-friendly probiotic since it can contribute to the host by reconstructing the balance of gut microbiota and only temporarily resides in the intestine after administration.W.coagulans has been generally recognized as safe(GRAS)by the US Food and Drug Administration(FDA),thus it is an ideal probiotic for improving gut health.The merit of its stability in processing and storage provides W.coagulans spores many possibilities for its use in various types of functional foods.This review presents an overview of the characteristics of W.coagulans that make it an ideal probiotic candidate and highlights the proposed health benefits with scientific evidence conferred by the administration of W.coagulans.
基金supported by Basic Science Research Program through the National Research Foundation of Korea(NRF)funded by the Korea Government(MSIT)(Grant No.NRF-2022R1A2C1004588).
文摘Reconstructing a three-dimensional(3D)environment is an indispensable technique to make augmented reality and augmented virtuality feasible.A Kinect device is an efficient tool for reconstructing 3D environments,and using multiple Kinect devices enables the enhancement of reconstruction density and expansion of virtual spaces.To employ multiple devices simultaneously,Kinect devices need to be calibrated with respect to each other.There are several schemes available that calibrate 3D images generated frommultiple Kinect devices,including themarker detection method.In this study,we introduce a markerless calibration technique for Azure Kinect devices that avoids the drawbacks of marker detection,which directly affects calibration accuracy;it offers superior userfriendliness,efficiency,and accuracy.Further,we applied a joint tracking algorithm to approximate the calibration.Traditional methods require the information of multiple joints for calibration;however,Azure Kinect,the latest version of Kinect,requires the information of only one joint.The obtained result was further refined using the iterative closest point algorithm.We conducted several experimental tests that confirmed the enhanced efficiency and accuracy of the proposed method for multiple Kinect devices when compared to the conventional markerbased calibration.
基金This work was supported by Institute of Information&communications Technology Planning&Evaluation(IITP)grant funded by the Korea Government(MSIT)(No.2021-0-00230,‘Development of real virtual environmental analysis based adaptive interaction technology’).
文摘Recently,simultaneous localization and mapping(SLAM)has received considerable attention in augmented reality(AR)libraries and applications.Although the assumption of scene rigidity is common in most visual SLAMs,this assumption limits the possibilities of AR applications in various real-world environments.In this paper,we propose a new tracking system that integrates SLAMwith amarker detection module for real-time AR applications in static and dynamic environments.Because the proposed system assumes that the marker is movable,SLAM performs tracking andmappingof the static scene except for themarker,andthemarker detector estimates the 3-dimensional pose of the marker attached to the dynamic object.We can place SLAM maps,cameras,and dynamic objects(markers)on the same coordinate system,and we can also accurately augment AR resources in real-time on both the static scene and dynamic objects simultaneously.Additionally,having a static map of the scene in advance has the advantage of being able to performtracking of static and dynamic environments with mapping disabled.In this paper,we evaluate whether the proposed tracking systemis suitable for AR applications and describe a strategy for creating AR applications in static and dynamic environments with a demonstration.
