Improved CenterNet for Accurate and Fast Fitting Object Detection
Accurate and fast detection of typical fittings is the prerequisite of condition monitoring and fault diagnosis. At present, most successful fitting detectors are anchor-based, which are challenging to meet the requirements of edge deployment. In this paper, we propose a novel anchor-free method cal...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2022-01-01
|
| Series: | Discrete Dynamics in Nature and Society |
| Online Access: | http://dx.doi.org/10.1155/2022/8417295 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832565761787297792 |
|---|---|
| author | Huimin He Qionglan Na Dan Su Kai Zhao Jing Lou Yixi Yang |
| author_facet | Huimin He Qionglan Na Dan Su Kai Zhao Jing Lou Yixi Yang |
| author_sort | Huimin He |
| collection | DOAJ |
| description | Accurate and fast detection of typical fittings is the prerequisite of condition monitoring and fault diagnosis. At present, most successful fitting detectors are anchor-based, which are challenging to meet the requirements of edge deployment. In this paper, we propose a novel anchor-free method called HRM-CenterNet. Firstly, the lightweight MobileNetV3 is introduced into CenterNet to extract multi-scale features of different layers. In addition, the lightweight receptive field enhancement module is proposed for the deep layer features to further enhance the characterization power of global features and generate more accurate heatmaps. Finally, the high-resolution feature fusion network with iterative aggregation is designed to reduce the loss of spatial semantic information in subsampling and further improve the accuracy of small and occlusion objects. Experiments are carried out on the TFITS and PASCAL VOC datasets. The results show that the size of the network is more than 60% lower than that of CenterNet. Compared with other detectors, our method achieves comparable accuracy with all accurate models at a much faster speed and meets the performance requirements of real-time detection. |
| format | Article |
| id | doaj-art-0a7c69249ece47d283caa202d8d5f642 |
| institution | Kabale University |
| issn | 1607-887X |
| language | English |
| publishDate | 2022-01-01 |
| publisher | Wiley |
| record_format | Article |
| series | Discrete Dynamics in Nature and Society |
| spelling | doaj-art-0a7c69249ece47d283caa202d8d5f6422025-02-03T01:06:36ZengWileyDiscrete Dynamics in Nature and Society1607-887X2022-01-01202210.1155/2022/8417295Improved CenterNet for Accurate and Fast Fitting Object DetectionHuimin He0Qionglan Na1Dan Su2Kai Zhao3Jing Lou4Yixi Yang5State Grid Jibei Information and Telecommunication CompanyState Grid Jibei Information and Telecommunication CompanyState Grid Jibei Information and Telecommunication CompanyNorth China Electric Power UniversityState Grid Jibei Information and Telecommunication CompanyState Grid Information and Telecommunication BranchAccurate and fast detection of typical fittings is the prerequisite of condition monitoring and fault diagnosis. At present, most successful fitting detectors are anchor-based, which are challenging to meet the requirements of edge deployment. In this paper, we propose a novel anchor-free method called HRM-CenterNet. Firstly, the lightweight MobileNetV3 is introduced into CenterNet to extract multi-scale features of different layers. In addition, the lightweight receptive field enhancement module is proposed for the deep layer features to further enhance the characterization power of global features and generate more accurate heatmaps. Finally, the high-resolution feature fusion network with iterative aggregation is designed to reduce the loss of spatial semantic information in subsampling and further improve the accuracy of small and occlusion objects. Experiments are carried out on the TFITS and PASCAL VOC datasets. The results show that the size of the network is more than 60% lower than that of CenterNet. Compared with other detectors, our method achieves comparable accuracy with all accurate models at a much faster speed and meets the performance requirements of real-time detection.http://dx.doi.org/10.1155/2022/8417295 |
| spellingShingle | Huimin He Qionglan Na Dan Su Kai Zhao Jing Lou Yixi Yang Improved CenterNet for Accurate and Fast Fitting Object Detection Discrete Dynamics in Nature and Society |
| title | Improved CenterNet for Accurate and Fast Fitting Object Detection |
| title_full | Improved CenterNet for Accurate and Fast Fitting Object Detection |
| title_fullStr | Improved CenterNet for Accurate and Fast Fitting Object Detection |
| title_full_unstemmed | Improved CenterNet for Accurate and Fast Fitting Object Detection |
| title_short | Improved CenterNet for Accurate and Fast Fitting Object Detection |
| title_sort | improved centernet for accurate and fast fitting object detection |
| url | http://dx.doi.org/10.1155/2022/8417295 |
| work_keys_str_mv | AT huiminhe improvedcenternetforaccurateandfastfittingobjectdetection AT qionglanna improvedcenternetforaccurateandfastfittingobjectdetection AT dansu improvedcenternetforaccurateandfastfittingobjectdetection AT kaizhao improvedcenternetforaccurateandfastfittingobjectdetection AT jinglou improvedcenternetforaccurateandfastfittingobjectdetection AT yixiyang improvedcenternetforaccurateandfastfittingobjectdetection |