SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks
Small object detection is one of the research difficulties in object detection, and Feature Pyramid Networks (FPN) is a common feature extractor in deep learning; thus, improving the results of small object detection based on FPN is of great significance in this field. In this paper, SV-FPN is propo...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-06-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/11/13/2028 |
| _version_ | 1797480393016868864 |
|---|---|
| author | Qianhui Yang Changlun Zhang Hengyou Wang Qiang He Lianzhi Huo |
| author_facet | Qianhui Yang Changlun Zhang Hengyou Wang Qiang He Lianzhi Huo |
| author_sort | Qianhui Yang |
| collection | DOAJ |
| description | Small object detection is one of the research difficulties in object detection, and Feature Pyramid Networks (FPN) is a common feature extractor in deep learning; thus, improving the results of small object detection based on FPN is of great significance in this field. In this paper, SV-FPN is proposed for a small object detection task, which consists of Small Object Feature Enhancement (SOFE) and Variance-guided Region of Interest Fusion (VRoIF). When using FPN as a feature extractor, an SOFE module is designed to enhance the finer-resolution level feature maps from which the small object features are extracted. VRoIF takes the variance of RoI features as the data driver to learn the completeness of several RoI features from different feature layers, which avoids wasting information and introducing noise. Ablation experiments on three public datasets (KITTI, PASCAL VOC 07+12 and MS COCO 2017) demonstrate the effectiveness of SV-FPN, and the mean Average Precision (mAP) of SV-FPN in the three datasets achieves 41.5%, 53.9% and 38.3%, respectively. |
| first_indexed | 2024-03-09T21:59:20Z |
| format | Article |
| id | doaj.art-22a872490d5446218a5ad1071a9f19fb |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-09T21:59:20Z |
| publishDate | 2022-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-22a872490d5446218a5ad1071a9f19fb2023-11-23T19:51:41ZengMDPI AGElectronics2079-92922022-06-011113202810.3390/electronics11132028SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid NetworksQianhui Yang0Changlun Zhang1Hengyou Wang2Qiang He3Lianzhi Huo4School of Science, Beijing University of Civil Engineering and Architecture, Beijing 100044, ChinaSchool of Science, Beijing University of Civil Engineering and Architecture, Beijing 100044, ChinaSchool of Science, Beijing University of Civil Engineering and Architecture, Beijing 100044, ChinaSchool of Science, Beijing University of Civil Engineering and Architecture, Beijing 100044, ChinaAerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, ChinaSmall object detection is one of the research difficulties in object detection, and Feature Pyramid Networks (FPN) is a common feature extractor in deep learning; thus, improving the results of small object detection based on FPN is of great significance in this field. In this paper, SV-FPN is proposed for a small object detection task, which consists of Small Object Feature Enhancement (SOFE) and Variance-guided Region of Interest Fusion (VRoIF). When using FPN as a feature extractor, an SOFE module is designed to enhance the finer-resolution level feature maps from which the small object features are extracted. VRoIF takes the variance of RoI features as the data driver to learn the completeness of several RoI features from different feature layers, which avoids wasting information and introducing noise. Ablation experiments on three public datasets (KITTI, PASCAL VOC 07+12 and MS COCO 2017) demonstrate the effectiveness of SV-FPN, and the mean Average Precision (mAP) of SV-FPN in the three datasets achieves 41.5%, 53.9% and 38.3%, respectively.https://www.mdpi.com/2079-9292/11/13/2028small object detectionFPNfeature enhancementRoI feature fusion |
| spellingShingle | Qianhui Yang Changlun Zhang Hengyou Wang Qiang He Lianzhi Huo SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks Electronics small object detection FPN feature enhancement RoI feature fusion |
| title | SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks |
| title_full | SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks |
| title_fullStr | SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks |
| title_full_unstemmed | SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks |
| title_short | SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks |
| title_sort | sv fpn small object feature enhancement and variance guided roi fusion for feature pyramid networks |
| topic | small object detection FPN feature enhancement RoI feature fusion |
| url | https://www.mdpi.com/2079-9292/11/13/2028 |
| work_keys_str_mv | AT qianhuiyang svfpnsmallobjectfeatureenhancementandvarianceguidedroifusionforfeaturepyramidnetworks AT changlunzhang svfpnsmallobjectfeatureenhancementandvarianceguidedroifusionforfeaturepyramidnetworks AT hengyouwang svfpnsmallobjectfeatureenhancementandvarianceguidedroifusionforfeaturepyramidnetworks AT qianghe svfpnsmallobjectfeatureenhancementandvarianceguidedroifusionforfeaturepyramidnetworks AT lianzhihuo svfpnsmallobjectfeatureenhancementandvarianceguidedroifusionforfeaturepyramidnetworks |