Person re-identification （re-id） on robot platform is an important application for human-robot- interaction （HRI）, which aims at making the robot recognize the around persons in varying scenes. Although many effec...Person re-identification （re-id） on robot platform is an important application for human-robot- interaction （HRI）, which aims at making the robot recognize the around persons in varying scenes. Although many effective methods have been proposed for surveillance re-id in recent years, re-id on robot platform is still a novel unsolved problem. Most existing methods adapt the supervised metric learning offline to improve the accuracy. However, these methods can not adapt to unknown scenes. To solve this problem, an online re-id framework is proposed. Considering that robotics can afford to use high-resolution RGB-D sensors and clear human face may be captured, face information is used to update the metric model. Firstly, the metric model is pre-trained offline using labeled data. Then during the online stage, we use face information to mine incorrect body matching pairs which are collected to update the metric model online. In addition, to make full use of both appearance and skeleton information provided by RGB-D sensors, a novel feature funnel model （FFM） is proposed. Comparison studies show our approach is more effective and adaptable to varying environments.展开更多
Safety accidents in the operation field of the distribution network often occur,which seriously endanger the safety and lives of operators.Existing identification methods for safety risk can identify static safety ris...Safety accidents in the operation field of the distribution network often occur,which seriously endanger the safety and lives of operators.Existing identification methods for safety risk can identify static safety risks,such as no-helmet,no-safety gloves,etc.,but fail to identify risks in the dynamic actions of operators.Therefore,this paper proposes a skeletonbased violation action-recognition method for supervision of safety during operations in a distribution network,i.e.,based on spatial temporal graph convolutional network(STGCN)and key joint attention module(KJAM),which can implement dynamic violation behavior recognition of operators.In this method,the human posture estimation method,i.e.Multi-Person Pose Estimation,is utilized to extract the skeleton information of operators during operations,and to construct an undirected graph,which reflects the movement and posture of the human body.Then,the STGCN is utilized to identify actions of operators that can lead to dynamic violations.In addition,the KJAM captures important joint information of operators.The effectiveness and superiority of the proposed method are verified in comparison to other action recognition methods.The experimental results show that the proposed method has higher recognition accuracy for common violations collected at the actual operation site of the distribution network and shows a strong generalization ability,which can be applied to the video monitoring system of field operations to reduce the occurrence of safety accidents.展开更多
基金This work is supported by the National Natural Science Foundation of China （NSFC, nos. 61340046）, the National High Technology Research and Development Programme of China （863 Programme, no. 2006AA04Z247）, the Scientific and Technical Innovation Commission of Shenzhen Municipality （nos. JCYJ20130331144631730）, and the Specialized Research Fund for the Doctoral Programme of Higher Education （SRFDP, no. 20130001110011）.
文摘Person re-identification （re-id） on robot platform is an important application for human-robot- interaction （HRI）, which aims at making the robot recognize the around persons in varying scenes. Although many effective methods have been proposed for surveillance re-id in recent years, re-id on robot platform is still a novel unsolved problem. Most existing methods adapt the supervised metric learning offline to improve the accuracy. However, these methods can not adapt to unknown scenes. To solve this problem, an online re-id framework is proposed. Considering that robotics can afford to use high-resolution RGB-D sensors and clear human face may be captured, face information is used to update the metric model. Firstly, the metric model is pre-trained offline using labeled data. Then during the online stage, we use face information to mine incorrect body matching pairs which are collected to update the metric model online. In addition, to make full use of both appearance and skeleton information provided by RGB-D sensors, a novel feature funnel model （FFM） is proposed. Comparison studies show our approach is more effective and adaptable to varying environments.
基金the Guizhou Province Science and Technology Plan Project(Gan ke he zhi cheng G.[2020]2Y039)the National Natural Science Foundation of China(No.51779206).
文摘Safety accidents in the operation field of the distribution network often occur,which seriously endanger the safety and lives of operators.Existing identification methods for safety risk can identify static safety risks,such as no-helmet,no-safety gloves,etc.,but fail to identify risks in the dynamic actions of operators.Therefore,this paper proposes a skeletonbased violation action-recognition method for supervision of safety during operations in a distribution network,i.e.,based on spatial temporal graph convolutional network(STGCN)and key joint attention module(KJAM),which can implement dynamic violation behavior recognition of operators.In this method,the human posture estimation method,i.e.Multi-Person Pose Estimation,is utilized to extract the skeleton information of operators during operations,and to construct an undirected graph,which reflects the movement and posture of the human body.Then,the STGCN is utilized to identify actions of operators that can lead to dynamic violations.In addition,the KJAM captures important joint information of operators.The effectiveness and superiority of the proposed method are verified in comparison to other action recognition methods.The experimental results show that the proposed method has higher recognition accuracy for common violations collected at the actual operation site of the distribution network and shows a strong generalization ability,which can be applied to the video monitoring system of field operations to reduce the occurrence of safety accidents.
