Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method
The monitoring video of underground power distribution room has a long duration and complex behavior types, and the traditional two-stream convolutional neural network (CNN) has poor recognition effect on such behaviors. In view of the problem, the two-stream CNN method was improved, and a method of...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Industry and Mine Automation
2020-04-01
|
| Series: | Gong-kuang zidonghua |
| Subjects: | |
| Online Access: | http://www.gkzdh.cn/article/doi/10.13272/j.issn.1671-251x.2019080074 |
| _version_ | 1818720731355152384 |
|---|---|
| author | DANG Weichao ZHANG Zejie BAI Shangwang GONG Dali WU Zhefeng |
| author_facet | DANG Weichao ZHANG Zejie BAI Shangwang GONG Dali WU Zhefeng |
| author_sort | DANG Weichao |
| collection | DOAJ |
| description | The monitoring video of underground power distribution room has a long duration and complex behavior types, and the traditional two-stream convolutional neural network (CNN) has poor recognition effect on such behaviors. In view of the problem, the two-stream CNN method was improved, and a method of inspection behavior recognition of underground power distribution room based on improved two-stream CNN was proposed. Through scene analysis, the inspection behaviors are divided into five types: standing detection, squatting detection, walking, standing record, and sitting down record, and the inspection behavior dataset IBDS5 is produced. Each inspection behavior video is divided into three parts, corresponding to the start of inspection, middle inspection and end of inspection; RGB images representing spatial features and continuous optical flow images representing motion features are obtained by random sample from three parts of the video, and the images are input to spatial flow network and time flow network respectively for feature extraction; weighted fusion of predicted features of the two networks are performed to obtain inspection behavior recognition results. The experimental results show that the spatial-temporal and dual-stream fusion network based on ResNet152 network structure with a weight ratio of 1∶2 has high recognition accuracy, and Top-1 accuracy reaches 98.92%;the recognition accuracy of the proposed method on the IBDS5 dataset and the public dataset UCF101 are better than existing methods such as 3D-CNN and traditional two-stream CNN. |
| first_indexed | 2024-12-17T20:27:30Z |
| format | Article |
| id | doaj.art-c59b975398ab4252b9f488b20dab5bbc |
| institution | Directory Open Access Journal |
| issn | 1671-251X |
| language | zho |
| last_indexed | 2024-12-17T20:27:30Z |
| publishDate | 2020-04-01 |
| publisher | Editorial Department of Industry and Mine Automation |
| record_format | Article |
| series | Gong-kuang zidonghua |
| spelling | doaj.art-c59b975398ab4252b9f488b20dab5bbc2022-12-21T21:33:43ZzhoEditorial Department of Industry and Mine AutomationGong-kuang zidonghua1671-251X2020-04-01464758010.13272/j.issn.1671-251x.2019080074Inspection behavior recognition of underground power distribution room based on improved two-stream CNN methodDANG WeichaoZHANG ZejieBAI ShangwangGONG DaliWU ZhefengThe monitoring video of underground power distribution room has a long duration and complex behavior types, and the traditional two-stream convolutional neural network (CNN) has poor recognition effect on such behaviors. In view of the problem, the two-stream CNN method was improved, and a method of inspection behavior recognition of underground power distribution room based on improved two-stream CNN was proposed. Through scene analysis, the inspection behaviors are divided into five types: standing detection, squatting detection, walking, standing record, and sitting down record, and the inspection behavior dataset IBDS5 is produced. Each inspection behavior video is divided into three parts, corresponding to the start of inspection, middle inspection and end of inspection; RGB images representing spatial features and continuous optical flow images representing motion features are obtained by random sample from three parts of the video, and the images are input to spatial flow network and time flow network respectively for feature extraction; weighted fusion of predicted features of the two networks are performed to obtain inspection behavior recognition results. The experimental results show that the spatial-temporal and dual-stream fusion network based on ResNet152 network structure with a weight ratio of 1∶2 has high recognition accuracy, and Top-1 accuracy reaches 98.92%;the recognition accuracy of the proposed method on the IBDS5 dataset and the public dataset UCF101 are better than existing methods such as 3D-CNN and traditional two-stream CNN.http://www.gkzdh.cn/article/doi/10.13272/j.issn.1671-251x.2019080074underground power distribution roominspection behavior recognitiontwo-stream cnnvideo segmentationfeature fusio |
| spellingShingle | DANG Weichao ZHANG Zejie BAI Shangwang GONG Dali WU Zhefeng Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method Gong-kuang zidonghua underground power distribution room inspection behavior recognition two-stream cnn video segmentation feature fusio |
| title | Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method |
| title_full | Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method |
| title_fullStr | Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method |
| title_full_unstemmed | Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method |
| title_short | Inspection behavior recognition of underground power distribution room based on improved two-stream CNN method |
| title_sort | inspection behavior recognition of underground power distribution room based on improved two stream cnn method |
| topic | underground power distribution room inspection behavior recognition two-stream cnn video segmentation feature fusio |
| url | http://www.gkzdh.cn/article/doi/10.13272/j.issn.1671-251x.2019080074 |
| work_keys_str_mv | AT dangweichao inspectionbehaviorrecognitionofundergroundpowerdistributionroombasedonimprovedtwostreamcnnmethod AT zhangzejie inspectionbehaviorrecognitionofundergroundpowerdistributionroombasedonimprovedtwostreamcnnmethod AT baishangwang inspectionbehaviorrecognitionofundergroundpowerdistributionroombasedonimprovedtwostreamcnnmethod AT gongdali inspectionbehaviorrecognitionofundergroundpowerdistributionroombasedonimprovedtwostreamcnnmethod AT wuzhefeng inspectionbehaviorrecognitionofundergroundpowerdistributionroombasedonimprovedtwostreamcnnmethod |