Cargando…
Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet
The existing pedestrian detection algorithms cannot effectively extract features of heavily occluded targets which results in lower detection accuracy. To solve the heavy occlusion in crowds, we propose a multi-scale feature pyramid network based on ResNet (MFPN) to enhance the features of occluded...
Autores principales: | , , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
MDPI
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7961544/ https://www.ncbi.nlm.nih.gov/pubmed/33807795 http://dx.doi.org/10.3390/s21051820 |
_version_ | 1783665283890675712 |
---|---|
author | Shao, Xiaotao Wang, Qing Yang, Wei Chen, Yun Xie, Yi Shen, Yan Wang, Zhongli |
author_facet | Shao, Xiaotao Wang, Qing Yang, Wei Chen, Yun Xie, Yi Shen, Yan Wang, Zhongli |
author_sort | Shao, Xiaotao |
collection | PubMed |
description | The existing pedestrian detection algorithms cannot effectively extract features of heavily occluded targets which results in lower detection accuracy. To solve the heavy occlusion in crowds, we propose a multi-scale feature pyramid network based on ResNet (MFPN) to enhance the features of occluded targets and improve the detection accuracy. MFPN includes two modules, namely double feature pyramid network (FPN) integrated with ResNet (DFR) and repulsion loss of minimum (RLM). We propose the double FPN which improves the architecture to further enhance the semantic information and contours of occluded pedestrians, and provide a new way for feature extraction of occluded targets. The features extracted by our network can be more separated and clearer, especially those heavily occluded pedestrians. Repulsion loss is introduced to improve the loss function which can keep predicted boxes away from the ground truths of the unrelated targets. Experiments carried out on the public CrowdHuman dataset, we obtain 90.96% AP which yields the best performance, 5.16% AP gains compared to the FPN-ResNet50 baseline. Compared with the state-of-the-art works, the performance of the pedestrian detection system has been boosted with our method. |
format | Online Article Text |
id | pubmed-7961544 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2021 |
publisher | MDPI |
record_format | MEDLINE/PubMed |
spelling | pubmed-79615442021-03-17 Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet Shao, Xiaotao Wang, Qing Yang, Wei Chen, Yun Xie, Yi Shen, Yan Wang, Zhongli Sensors (Basel) Article The existing pedestrian detection algorithms cannot effectively extract features of heavily occluded targets which results in lower detection accuracy. To solve the heavy occlusion in crowds, we propose a multi-scale feature pyramid network based on ResNet (MFPN) to enhance the features of occluded targets and improve the detection accuracy. MFPN includes two modules, namely double feature pyramid network (FPN) integrated with ResNet (DFR) and repulsion loss of minimum (RLM). We propose the double FPN which improves the architecture to further enhance the semantic information and contours of occluded pedestrians, and provide a new way for feature extraction of occluded targets. The features extracted by our network can be more separated and clearer, especially those heavily occluded pedestrians. Repulsion loss is introduced to improve the loss function which can keep predicted boxes away from the ground truths of the unrelated targets. Experiments carried out on the public CrowdHuman dataset, we obtain 90.96% AP which yields the best performance, 5.16% AP gains compared to the FPN-ResNet50 baseline. Compared with the state-of-the-art works, the performance of the pedestrian detection system has been boosted with our method. MDPI 2021-03-05 /pmc/articles/PMC7961544/ /pubmed/33807795 http://dx.doi.org/10.3390/s21051820 Text en © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). |
spellingShingle | Article Shao, Xiaotao Wang, Qing Yang, Wei Chen, Yun Xie, Yi Shen, Yan Wang, Zhongli Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title | Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title_full | Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title_fullStr | Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title_full_unstemmed | Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title_short | Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet |
title_sort | multi-scale feature pyramid network: a heavily occluded pedestrian detection network based on resnet |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7961544/ https://www.ncbi.nlm.nih.gov/pubmed/33807795 http://dx.doi.org/10.3390/s21051820 |
work_keys_str_mv | AT shaoxiaotao multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT wangqing multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT yangwei multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT chenyun multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT xieyi multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT shenyan multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet AT wangzhongli multiscalefeaturepyramidnetworkaheavilyoccludedpedestriandetectionnetworkbasedonresnet |