Rock Particle Motion Information Detection Based on Video Instance Segmentation
The detection of rock particle motion information is the basis for revealing particle motion laws and quantitative analysis. Such a task is crucial in guiding engineering construction, preventing geological disasters, and verifying numerical models of particles. We propose a machine vision method ba...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-06-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/21/12/4108 |
| _version_ | 1797530099081281536 |
|---|---|
| author | Man Chen Maojun Li Yiwei Li Wukun Yi |
| author_facet | Man Chen Maojun Li Yiwei Li Wukun Yi |
| author_sort | Man Chen |
| collection | DOAJ |
| description | The detection of rock particle motion information is the basis for revealing particle motion laws and quantitative analysis. Such a task is crucial in guiding engineering construction, preventing geological disasters, and verifying numerical models of particles. We propose a machine vision method based on video instance segmentation (VIS) to address the motion information detection problem in rock particles under a vibration load. First, we designed a classification loss function based on Arcface loss to improve the Mask R-CNN. This loss function introduces an angular distance based on SoftMax loss that distinguishes the objects and backgrounds with higher similarity. Second, this method combines the abovementioned Mask R-CNN and Deep Simple Online and Real-time Tracking (Deep SORT) to perform rock particle detection, segmentation, and tracking. Third, we utilized the equivalent ellipse characterization method for segmented particles, integrating with the proportional calibration algorithm to test the translation and detecting the rotation by calculating the change in the angle of the ellipse’s major axis. The experimental results show that the improved Mask R-CNN obtains an accuracy of 93.36% on a self-created dataset and also has some advantages on public datasets. Combining the improved Mask R-CNN and Deep SORT could fulfill the VIS with a low ID switching rate while successfully detecting movement information. The average detection errors of translation and rotation are 5.10% and 14.49%, respectively. This study provides an intelligent scheme for detecting movement information of rock particles. |
| first_indexed | 2024-03-10T10:24:07Z |
| format | Article |
| id | doaj.art-1f33c175c07a4d458e25aaf83ef31bb4 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T10:24:07Z |
| publishDate | 2021-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-1f33c175c07a4d458e25aaf83ef31bb42023-11-22T00:09:48ZengMDPI AGSensors1424-82202021-06-012112410810.3390/s21124108Rock Particle Motion Information Detection Based on Video Instance SegmentationMan Chen0Maojun Li1Yiwei Li2Wukun Yi3School of Electrical and Information Engineering, Changsha University of Science and Technology, Changsha 410114, ChinaSchool of Electrical and Information Engineering, Changsha University of Science and Technology, Changsha 410114, ChinaSchool of Electrical and Information Engineering, Changsha University of Science and Technology, Changsha 410114, ChinaXiaoxiang Research Institute of Big Data, Changsha 410199, ChinaThe detection of rock particle motion information is the basis for revealing particle motion laws and quantitative analysis. Such a task is crucial in guiding engineering construction, preventing geological disasters, and verifying numerical models of particles. We propose a machine vision method based on video instance segmentation (VIS) to address the motion information detection problem in rock particles under a vibration load. First, we designed a classification loss function based on Arcface loss to improve the Mask R-CNN. This loss function introduces an angular distance based on SoftMax loss that distinguishes the objects and backgrounds with higher similarity. Second, this method combines the abovementioned Mask R-CNN and Deep Simple Online and Real-time Tracking (Deep SORT) to perform rock particle detection, segmentation, and tracking. Third, we utilized the equivalent ellipse characterization method for segmented particles, integrating with the proportional calibration algorithm to test the translation and detecting the rotation by calculating the change in the angle of the ellipse’s major axis. The experimental results show that the improved Mask R-CNN obtains an accuracy of 93.36% on a self-created dataset and also has some advantages on public datasets. Combining the improved Mask R-CNN and Deep SORT could fulfill the VIS with a low ID switching rate while successfully detecting movement information. The average detection errors of translation and rotation are 5.10% and 14.49%, respectively. This study provides an intelligent scheme for detecting movement information of rock particles.https://www.mdpi.com/1424-8220/21/12/4108rock particlesmachine visionvideo instance segmentationmotion information detection |
| spellingShingle | Man Chen Maojun Li Yiwei Li Wukun Yi Rock Particle Motion Information Detection Based on Video Instance Segmentation Sensors rock particles machine vision video instance segmentation motion information detection |
| title | Rock Particle Motion Information Detection Based on Video Instance Segmentation |
| title_full | Rock Particle Motion Information Detection Based on Video Instance Segmentation |
| title_fullStr | Rock Particle Motion Information Detection Based on Video Instance Segmentation |
| title_full_unstemmed | Rock Particle Motion Information Detection Based on Video Instance Segmentation |
| title_short | Rock Particle Motion Information Detection Based on Video Instance Segmentation |
| title_sort | rock particle motion information detection based on video instance segmentation |
| topic | rock particles machine vision video instance segmentation motion information detection |
| url | https://www.mdpi.com/1424-8220/21/12/4108 |
| work_keys_str_mv | AT manchen rockparticlemotioninformationdetectionbasedonvideoinstancesegmentation AT maojunli rockparticlemotioninformationdetectionbasedonvideoinstancesegmentation AT yiweili rockparticlemotioninformationdetectionbasedonvideoinstancesegmentation AT wukunyi rockparticlemotioninformationdetectionbasedonvideoinstancesegmentation |