The precise segmentation of point cloud is the basic work of 3D reconstruction, which has important value in scene understanding fields such as unmanned driving, crop phenotype measurement, fruit grading, picking robots and postharvest sorting. At present, there are many effective remedies to improve the speed and accuracy of segmentation on simple scenes. However, due to the complexity of real-world and agricultural environment and the limitations of 3D acquisition devices, it is still challenging to efficiently segment the target object from the chaotic real background. In this paper, we provide a comprehensive review of point cloud segmentation techniques, and classify point cloud segmentation methods based on different mathematical principles to provide insight to inspire further research. To address the problem that point cloud segmentation algorithms are difficult to evaluate, a quantitative evaluation scheme is proposed by combining segmentation precision, recall and F-score. Finally, we analyze the advantages and disadvantages of algorithms for application in the agricultural field through fruit point cloud segmentation experiments. At the same time, potential future research directions is inspired in this rapidly developing field.
{"title":"Research on Crop Fruit Segmentation Method Based on Point Cloud","authors":"Ruiping Wang, Danni Yang, Yuan Ma, Dongfeng Liu, Xin Wang, Hui-jun Yang","doi":"10.1109/ICVR57957.2023.10169303","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169303","url":null,"abstract":"The precise segmentation of point cloud is the basic work of 3D reconstruction, which has important value in scene understanding fields such as unmanned driving, crop phenotype measurement, fruit grading, picking robots and postharvest sorting. At present, there are many effective remedies to improve the speed and accuracy of segmentation on simple scenes. However, due to the complexity of real-world and agricultural environment and the limitations of 3D acquisition devices, it is still challenging to efficiently segment the target object from the chaotic real background. In this paper, we provide a comprehensive review of point cloud segmentation techniques, and classify point cloud segmentation methods based on different mathematical principles to provide insight to inspire further research. To address the problem that point cloud segmentation algorithms are difficult to evaluate, a quantitative evaluation scheme is proposed by combining segmentation precision, recall and F-score. Finally, we analyze the advantages and disadvantages of algorithms for application in the agricultural field through fruit point cloud segmentation experiments. At the same time, potential future research directions is inspired in this rapidly developing field.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123981161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169218
Junjie Fu, R. Liu, Pengfei Yi, Jing Dong, D. Zhou, Xiaopeng Wei
Pedestrian search is a combined task of pedestrian detection and pedestrian re-identification, i.e., how to find a specific pedestrian in a panoramic picture. However, due to the impact of different imaging conditions and environments, the accuracy of pedestrian search is still low at present, which is difficult to meet the needs of practical applications. To improve the accuracy of pedestrian search, this paper proposes PSA-CLNet pedestrian search method, which integrates polarized self-attention mechanism (PSA) and cosine online instance matching (COIM) loss. Firstly, the channel attention is added to enhance the pedestrian feature information and improve the expression ability of the feature information. Secondly, spatial attention is added to enhance the areas that have more contribution to the pedestrian search task to increase the correlation between the pictures and improve the recognition ability. Meanwhile, this paper proposes a COIM loss based on the online instance matching (OIM) loss. The loss function can enlarge the class distance between pedestrians and background, and reduce the class distance between pedestrians. This could improve the accuracy of pedestrian search algorithm effectively. Finally, this paper uses ResNet as the main backbone network to verify the proposed method. The pedestrian search method proposed in this paper could achieve 96.60% of mAP index and 96.83% of top-1 index in CUHK-SYSU dataset, and achieve 48.78% of mAP index and 88.38% of top-1 index in PRW dataset.
{"title":"PSA-CLNet: Pedestrian Search Method Based on Polarized Self-Attention and COIM Loss","authors":"Junjie Fu, R. Liu, Pengfei Yi, Jing Dong, D. Zhou, Xiaopeng Wei","doi":"10.1109/ICVR57957.2023.10169218","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169218","url":null,"abstract":"Pedestrian search is a combined task of pedestrian detection and pedestrian re-identification, i.e., how to find a specific pedestrian in a panoramic picture. However, due to the impact of different imaging conditions and environments, the accuracy of pedestrian search is still low at present, which is difficult to meet the needs of practical applications. To improve the accuracy of pedestrian search, this paper proposes PSA-CLNet pedestrian search method, which integrates polarized self-attention mechanism (PSA) and cosine online instance matching (COIM) loss. Firstly, the channel attention is added to enhance the pedestrian feature information and improve the expression ability of the feature information. Secondly, spatial attention is added to enhance the areas that have more contribution to the pedestrian search task to increase the correlation between the pictures and improve the recognition ability. Meanwhile, this paper proposes a COIM loss based on the online instance matching (OIM) loss. The loss function can enlarge the class distance between pedestrians and background, and reduce the class distance between pedestrians. This could improve the accuracy of pedestrian search algorithm effectively. Finally, this paper uses ResNet as the main backbone network to verify the proposed method. The pedestrian search method proposed in this paper could achieve 96.60% of mAP index and 96.83% of top-1 index in CUHK-SYSU dataset, and achieve 48.78% of mAP index and 88.38% of top-1 index in PRW dataset.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115529052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169153
Yi Liang, Yi Xiao, Guokai Feng, J. Shell, Conor Fahy, Yongkang Xing
Law students (Undergraduate) usually use a moot court on-campus to practice individual legal skills before they start their careers. The study addresses the limited availability and high cost of moot court facilities for law students in Guangdong’s universities. The study proposes utilizing virtual reality (VR) technology to solve the current issue. The VR solution aims to enhance the students’ legal skills through a virtual moot court system designed to mimic a real-life scenario. The system comprises a Case Study Mode that enables law students to practice classic cases and a Visiting Mode that allows non-law students to explore the VR moot court and acquire basic legal knowledge. Through an interactive and seamless learning experience, this study demonstrates the scalability and potential of VR technology in online education for law students. The design of this system was based on serious gaming theory and previous VR educational projects, including those outside the legal field. The results suggest that VR technology has the potential to benefit practical courses by attracting students through its realistic graphics and interactive elements.
{"title":"Towards a VR Moot Court for Law Students in Enhancing Practice Experience","authors":"Yi Liang, Yi Xiao, Guokai Feng, J. Shell, Conor Fahy, Yongkang Xing","doi":"10.1109/ICVR57957.2023.10169153","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169153","url":null,"abstract":"Law students (Undergraduate) usually use a moot court on-campus to practice individual legal skills before they start their careers. The study addresses the limited availability and high cost of moot court facilities for law students in Guangdong’s universities. The study proposes utilizing virtual reality (VR) technology to solve the current issue. The VR solution aims to enhance the students’ legal skills through a virtual moot court system designed to mimic a real-life scenario. The system comprises a Case Study Mode that enables law students to practice classic cases and a Visiting Mode that allows non-law students to explore the VR moot court and acquire basic legal knowledge. Through an interactive and seamless learning experience, this study demonstrates the scalability and potential of VR technology in online education for law students. The design of this system was based on serious gaming theory and previous VR educational projects, including those outside the legal field. The results suggest that VR technology has the potential to benefit practical courses by attracting students through its realistic graphics and interactive elements.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126530175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Research based on electroencephalogram (EEG) and virtual reality (VR) has developed rapidly in recent years. The integration of EEG and VR technology has been broadly employed in psychological intervention to treat various psychological conditions and alleviate stress. At present, there exist commercial-level EEG emotion recognition products, however, research on the correlation between the content of VR scenarios and emotions remains unavailable. Therefore, this work investigates VR scene construction based on EEG, which incorporates the recovery environment theory and virtual reality technology to construct different three-dimensional dynamic virtual reality interaction scenes. EEG equipment constructs an emotion-adaptive VR system based on EEG signals with the aid of VR interaction equipment. Herein, EEG data feedback was applied to automatically generate VR scenes, and the experimental findings revealed that the system has certain feasibility and application value.
{"title":"EEG-Based VR Scene Adaptive Generation System for Regulating Emotion","authors":"Hui Liang, Shiqing Liu, Yi Wang, Junjun Pan, Jialin Fu, Yingkai Yuan","doi":"10.1109/ICVR57957.2023.10169325","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169325","url":null,"abstract":"Research based on electroencephalogram (EEG) and virtual reality (VR) has developed rapidly in recent years. The integration of EEG and VR technology has been broadly employed in psychological intervention to treat various psychological conditions and alleviate stress. At present, there exist commercial-level EEG emotion recognition products, however, research on the correlation between the content of VR scenarios and emotions remains unavailable. Therefore, this work investigates VR scene construction based on EEG, which incorporates the recovery environment theory and virtual reality technology to construct different three-dimensional dynamic virtual reality interaction scenes. EEG equipment constructs an emotion-adaptive VR system based on EEG signals with the aid of VR interaction equipment. Herein, EEG data feedback was applied to automatically generate VR scenes, and the experimental findings revealed that the system has certain feasibility and application value.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129121237","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169498
Chen Chen, Di Wang
Tree modelling is a necessary topic for discussion in virtual world construction. This paper presents a new approach that uses point cloud reconstructions of the resulting tree skeleton to form a compact parametric L-system expression after branching parameter abstraction. The proposed strategy can generate diverse new parametric L-system expressions under the control of several iterations, allowing the construction of a set of tree models with diverse growth patterns but more similar to real trees. This technique makes it possible to create large-scale tree models in virtual environments.
{"title":"3D Tree Modeling Based on Abstract Parametric L-System","authors":"Chen Chen, Di Wang","doi":"10.1109/ICVR57957.2023.10169498","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169498","url":null,"abstract":"Tree modelling is a necessary topic for discussion in virtual world construction. This paper presents a new approach that uses point cloud reconstructions of the resulting tree skeleton to form a compact parametric L-system expression after branching parameter abstraction. The proposed strategy can generate diverse new parametric L-system expressions under the control of several iterations, allowing the construction of a set of tree models with diverse growth patterns but more similar to real trees. This technique makes it possible to create large-scale tree models in virtual environments.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127683021","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169645
Su Xin-yue, Xue Hao-Wei, Wang Mei-Li
Virtua1 reality (VR) technology has become one of the research hotspots of technology to promote teaching because of its high immersion and good interactivity. Putting VR games into education can make learning more participatory and incentive. Nevertheless, the majority of current technologies and efforts to accomplish human-computer interaction necessitate additional hardware, which has certain limits. To solve these problems, realize real-time operation to provide direct feedback, and enhance the user’s sense of immersion and experience, we have developed a real-time hand capture algorithm using a monocular camera to perform more accurate user gesture recognition. In the algorithm, we use the backbone architecture of deep residual network ResNet-50 as a feature extractor. Through the combined training of 2D and 3D annotation data, we are able to effectively predict 2D posture and 3D spatial information and implement virtual content interaction. This method achieves real-time performance (90fps) and accuracy (95.6%) on existing datasets and outperforms existing methods in hand mesh/pose accuracy and hand image alignment. We built and implemented a virtual reality game based on the method proposed in this research, and then transplanted it onto the VR platform with an ecological setting. While providing users with an immersive experience, we also want to use virtual reality technology to teach, play, and promote traditional Chinese culture.
{"title":"A 3D Hand Joint Detection Network for Real-Time Hand Capture and Its Application in a Game of Moving Mountains","authors":"Su Xin-yue, Xue Hao-Wei, Wang Mei-Li","doi":"10.1109/ICVR57957.2023.10169645","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169645","url":null,"abstract":"Virtua1 reality (VR) technology has become one of the research hotspots of technology to promote teaching because of its high immersion and good interactivity. Putting VR games into education can make learning more participatory and incentive. Nevertheless, the majority of current technologies and efforts to accomplish human-computer interaction necessitate additional hardware, which has certain limits. To solve these problems, realize real-time operation to provide direct feedback, and enhance the user’s sense of immersion and experience, we have developed a real-time hand capture algorithm using a monocular camera to perform more accurate user gesture recognition. In the algorithm, we use the backbone architecture of deep residual network ResNet-50 as a feature extractor. Through the combined training of 2D and 3D annotation data, we are able to effectively predict 2D posture and 3D spatial information and implement virtual content interaction. This method achieves real-time performance (90fps) and accuracy (95.6%) on existing datasets and outperforms existing methods in hand mesh/pose accuracy and hand image alignment. We built and implemented a virtual reality game based on the method proposed in this research, and then transplanted it onto the VR platform with an ecological setting. While providing users with an immersive experience, we also want to use virtual reality technology to teach, play, and promote traditional Chinese culture.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126487707","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169529
Chuan Yun, H. Su, Endong Han
With the increasing improvement of rail transit, the safety of railway traffic has received more and more attention, and efficient and accurate communication between train attendants is the basis of driving safety. Aiming at the gesture training of train attendants, this paper uses the MediaPipe module as the skeleton key point recognition framework on the basis of virtual reality technology, and designs a gesture recognition solution for train conductors by using skeleton information extraction technology and deep learning related methods. It can train train the conductor’s gestures, which has the characteristics of real-time, good interaction, initiative and flexibility.
{"title":"Motion Recognition System in VR Training System for Train Attendants","authors":"Chuan Yun, H. Su, Endong Han","doi":"10.1109/ICVR57957.2023.10169529","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169529","url":null,"abstract":"With the increasing improvement of rail transit, the safety of railway traffic has received more and more attention, and efficient and accurate communication between train attendants is the basis of driving safety. Aiming at the gesture training of train attendants, this paper uses the MediaPipe module as the skeleton key point recognition framework on the basis of virtual reality technology, and designs a gesture recognition solution for train conductors by using skeleton information extraction technology and deep learning related methods. It can train train the conductor’s gestures, which has the characteristics of real-time, good interaction, initiative and flexibility.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128445939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169333
Tilemachos K. Koliopoulos, G. Mratskova
This paper analyses the virtual reality simulation training useful digital monitoring utilities in terms of environmental impact assessment, sustainable development, safe health tourism construction infrastructures, agricultural tourism facilities at post COVID – 19 era and public health protection from alternative types of tourism for disable people and elderly. Useful analysis is taken into account based on useful digital virtual reality 3D animation utilities and multilingual digital drawings content for digital training, teaching and entertainment. In this paper are presented useful health policy outcomes for stakeholders, safe training schemes, digital drawings support for good operational management at sustainable designs and proper monitoring schemes for public and community health protection. Useful simulation e-learning tools are resented for stakeholders and training schemes applying efficient economic lightweight robust designs. Integrated health policy is presented within useful digital training e-learning simulation utilities and management techniques for monitoring environmental impacts like landfill emissions from renewable energy projects so as to protect public health. Useful results are presented for the safety of particular community health tourism units, and safe agriculture tourism facilities including interactive sports events for physical activities, applying proper designs for all and sustainable tourism infrastructures with safe ergonomic mobility facilities at post COVID-19 era.
{"title":"Digital Training, Teaching, Entertainment Utility for Disable and Elderly at Community Health & Agricultural Tourism Infrastructures","authors":"Tilemachos K. Koliopoulos, G. Mratskova","doi":"10.1109/ICVR57957.2023.10169333","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169333","url":null,"abstract":"This paper analyses the virtual reality simulation training useful digital monitoring utilities in terms of environmental impact assessment, sustainable development, safe health tourism construction infrastructures, agricultural tourism facilities at post COVID – 19 era and public health protection from alternative types of tourism for disable people and elderly. Useful analysis is taken into account based on useful digital virtual reality 3D animation utilities and multilingual digital drawings content for digital training, teaching and entertainment. In this paper are presented useful health policy outcomes for stakeholders, safe training schemes, digital drawings support for good operational management at sustainable designs and proper monitoring schemes for public and community health protection. Useful simulation e-learning tools are resented for stakeholders and training schemes applying efficient economic lightweight robust designs. Integrated health policy is presented within useful digital training e-learning simulation utilities and management techniques for monitoring environmental impacts like landfill emissions from renewable energy projects so as to protect public health. Useful results are presented for the safety of particular community health tourism units, and safe agriculture tourism facilities including interactive sports events for physical activities, applying proper designs for all and sustainable tourism infrastructures with safe ergonomic mobility facilities at post COVID-19 era.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125602162","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Most SLAM systems are based on the assumption of a static environment, while most of the real scenes with dynamic will cause mismatching in the process of camera pose estimation and affect localization accuracy and system robustness. To improve the effect and accuracy of SLAM dense reconstruction in dynamic environments and to increase the real-time performance of advanced tasks such as navigation and human-computer interaction, this paper proposes a real-time SLAM dense reconstruction method based on object detection in dynamic environments. Firstly, we use YOLOv5 for frame-by-frame detection and semantic marking of images to accurately identify the image feature points under dynamic semantic masks. Further, this paper proposed a keyframe filtering algorithm based on dynamic semantic marking to ensure the accuracy of dense map reconstruction by eliminating frames containing dynamic objects, which solves the interference of point cloud redundancy and moving objects in dynamic environments. Finally, this paper takes the scene depth information as an important reference for characterizing the geometric structure of the scene. By introducing the depth factor into the traditional feature extraction algorithm, we explored a new joint-depth-information based feature extraction and feature descriptor calculation method, and proposed a high precision visual SLAM based on deep joint visual odometer. Experiments on the TUM RGBD public dataset show that the trajectory accuracy of constructing dense point cloud maps under partial dynamic scene sequences is improved by 16.1% over DS-SLAM and 82.3% over ORB-SLAM2.
{"title":"High Precision ORB-SLAM Dense Reconstruction Based on Depth Visual Odometer in Dynamic Environments","authors":"Yinbing Chen, Hui-jun Yang, Jiajun Lu, Boxuan Jiang","doi":"10.1109/ICVR57957.2023.10169708","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169708","url":null,"abstract":"Most SLAM systems are based on the assumption of a static environment, while most of the real scenes with dynamic will cause mismatching in the process of camera pose estimation and affect localization accuracy and system robustness. To improve the effect and accuracy of SLAM dense reconstruction in dynamic environments and to increase the real-time performance of advanced tasks such as navigation and human-computer interaction, this paper proposes a real-time SLAM dense reconstruction method based on object detection in dynamic environments. Firstly, we use YOLOv5 for frame-by-frame detection and semantic marking of images to accurately identify the image feature points under dynamic semantic masks. Further, this paper proposed a keyframe filtering algorithm based on dynamic semantic marking to ensure the accuracy of dense map reconstruction by eliminating frames containing dynamic objects, which solves the interference of point cloud redundancy and moving objects in dynamic environments. Finally, this paper takes the scene depth information as an important reference for characterizing the geometric structure of the scene. By introducing the depth factor into the traditional feature extraction algorithm, we explored a new joint-depth-information based feature extraction and feature descriptor calculation method, and proposed a high precision visual SLAM based on deep joint visual odometer. Experiments on the TUM RGBD public dataset show that the trajectory accuracy of constructing dense point cloud maps under partial dynamic scene sequences is improved by 16.1% over DS-SLAM and 82.3% over ORB-SLAM2.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133766411","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/ICVR57957.2023.10169840
Jinfang Li, Mingtong He, Jiancong Su, Boyang Wang, Zhenxian Li
To meet the needs of teaching and practical applications in machine vision technology, a virtual reality-based machine vision experimental platform has been designed and developed. Unity3D was utilized as the development engine, and image processing technology was integrated to achieve the construction of virtual production line scenes, simulation of vision component parameter adjustments, and image acquisition. The platform features a graphical programming interface for visualizing image processing algorithms, which can be used to perform visual debugging of vision stations with a virtual robot system driven by software PLC. This machine vision experimental platform ensures the consistency between simulation and actual engineering processes, and enables students to explore different vision schemes on an industrial production line, thereby avoiding constraints on location, time, and equipment in related experiments.
{"title":"Design and Implementation of Machine Vision Experiment Platform for Virtual Production Line","authors":"Jinfang Li, Mingtong He, Jiancong Su, Boyang Wang, Zhenxian Li","doi":"10.1109/ICVR57957.2023.10169840","DOIUrl":"https://doi.org/10.1109/ICVR57957.2023.10169840","url":null,"abstract":"To meet the needs of teaching and practical applications in machine vision technology, a virtual reality-based machine vision experimental platform has been designed and developed. Unity3D was utilized as the development engine, and image processing technology was integrated to achieve the construction of virtual production line scenes, simulation of vision component parameter adjustments, and image acquisition. The platform features a graphical programming interface for visualizing image processing algorithms, which can be used to perform visual debugging of vision stations with a virtual robot system driven by software PLC. This machine vision experimental platform ensures the consistency between simulation and actual engineering processes, and enables students to explore different vision schemes on an industrial production line, thereby avoiding constraints on location, time, and equipment in related experiments.","PeriodicalId":439483,"journal":{"name":"2023 9th International Conference on Virtual Reality (ICVR)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133044710","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：