Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module
Pedestrian detection for complex scenes suffers from pedestrian occlusion issues, such as occlusions between pedestrians. As well-known, compared with the variability of the human body, the shape of a human head and their shoulders changes minimally and has high stability. Therefore, head detection...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-06-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/10/13/1565 |
| _version_ | 1797528425318055936 |
|---|---|
| author | Junwen Liu Yongjun Zhang Jianbin Xie Yan Wei Zewei Wang Mengjia Niu |
| author_facet | Junwen Liu Yongjun Zhang Jianbin Xie Yan Wei Zewei Wang Mengjia Niu |
| author_sort | Junwen Liu |
| collection | DOAJ |
| description | Pedestrian detection for complex scenes suffers from pedestrian occlusion issues, such as occlusions between pedestrians. As well-known, compared with the variability of the human body, the shape of a human head and their shoulders changes minimally and has high stability. Therefore, head detection is an important research area in the field of pedestrian detection. The translational invariance of neural network enables us to design a deep convolutional neural network, which means that, even if the appearance and location of the target changes, it can still be recognized effectively. However, the problems of scale invariance and high miss detection rates for small targets still exist. In this paper, a feature extraction network DR-Net based on Darknet-53 is proposed to improve the information transmission rate between convolutional layers and to extract more semantic information. In addition, the MDC (mixed dilated convolution) with different sampling rates of dilated convolution is embedded to improve the detection rate of small targets. We evaluated our method on three publicly available datasets and achieved excellent results. The AP (Average Precision) value on the Brainwash dataset, HollywoodHeads dataset, and SCUT-HEAD dataset reached 92.1%, 84.8%, and 90% respectively. |
| first_indexed | 2024-03-10T09:58:09Z |
| format | Article |
| id | doaj.art-dadff925344f424898711aa4b6d61c02 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T09:58:09Z |
| publishDate | 2021-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-dadff925344f424898711aa4b6d61c022023-11-22T02:10:44ZengMDPI AGElectronics2079-92922021-06-011013156510.3390/electronics10131565Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution ModuleJunwen Liu0Yongjun Zhang1Jianbin Xie2Yan Wei3Zewei Wang4Mengjia Niu5Key Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province, College of Computer Science and Technology, Guizhou University, Guiyang 550000, ChinaKey Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province, College of Computer Science and Technology, Guizhou University, Guiyang 550000, ChinaSchool of Electronic Science, National University of Defense Technology, Changsha 410073, ChinaSchool of Electronic Science, National University of Defense Technology, Changsha 410073, ChinaKey Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province, College of Computer Science and Technology, Guizhou University, Guiyang 550000, ChinaKey Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province, College of Computer Science and Technology, Guizhou University, Guiyang 550000, ChinaPedestrian detection for complex scenes suffers from pedestrian occlusion issues, such as occlusions between pedestrians. As well-known, compared with the variability of the human body, the shape of a human head and their shoulders changes minimally and has high stability. Therefore, head detection is an important research area in the field of pedestrian detection. The translational invariance of neural network enables us to design a deep convolutional neural network, which means that, even if the appearance and location of the target changes, it can still be recognized effectively. However, the problems of scale invariance and high miss detection rates for small targets still exist. In this paper, a feature extraction network DR-Net based on Darknet-53 is proposed to improve the information transmission rate between convolutional layers and to extract more semantic information. In addition, the MDC (mixed dilated convolution) with different sampling rates of dilated convolution is embedded to improve the detection rate of small targets. We evaluated our method on three publicly available datasets and achieved excellent results. The AP (Average Precision) value on the Brainwash dataset, HollywoodHeads dataset, and SCUT-HEAD dataset reached 92.1%, 84.8%, and 90% respectively.https://www.mdpi.com/2079-9292/10/13/1565head detectionsmall targetsconvolutional neural networksDR-NetMDC |
| spellingShingle | Junwen Liu Yongjun Zhang Jianbin Xie Yan Wei Zewei Wang Mengjia Niu Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module Electronics head detection small targets convolutional neural networks DR-Net MDC |
| title | Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module |
| title_full | Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module |
| title_fullStr | Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module |
| title_full_unstemmed | Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module |
| title_short | Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module |
| title_sort | head detection based on dr feature extraction network and mixed dilated convolution module |
| topic | head detection small targets convolutional neural networks DR-Net MDC |
| url | https://www.mdpi.com/2079-9292/10/13/1565 |
| work_keys_str_mv | AT junwenliu headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule AT yongjunzhang headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule AT jianbinxie headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule AT yanwei headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule AT zeweiwang headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule AT mengjianiu headdetectionbasedondrfeatureextractionnetworkandmixeddilatedconvolutionmodule |