A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network
In recent years, significant progress has been obtained in object detection using Convolutional Neural Networks (CNNs). However, owing to the particularity of Remote Sensing Images (RSIs), common object detection methods are not well suited for RSIs. Aiming at the difficulties in RSIs, this paper pr...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-03-01
|
| Series: | ISPRS International Journal of Geo-Information |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2220-9964/11/3/189 |
| _version_ | 1797471084128239616 |
|---|---|
| author | Liming Zhou Xiaohan Rao Yahui Li Xianyu Zuo Baojun Qiao Yinghao Lin |
| author_facet | Liming Zhou Xiaohan Rao Yahui Li Xianyu Zuo Baojun Qiao Yinghao Lin |
| author_sort | Liming Zhou |
| collection | DOAJ |
| description | In recent years, significant progress has been obtained in object detection using Convolutional Neural Networks (CNNs). However, owing to the particularity of Remote Sensing Images (RSIs), common object detection methods are not well suited for RSIs. Aiming at the difficulties in RSIs, this paper proposes an object detection method based on the Dense Feature Fusion Path Aggregation Network (DFF-PANet). Firstly, for better improving the detection performance of small and medium-sized instances, we propose Feature Reuse Module (FRM), which can integrate semantic and location information contained in feature maps; this module can reuse feature maps in the backbone to enhance the detection capability of small and medium-sized instances. After that, we design the DFF-PANet, which can help feature information extracted from the backbone to be fused more efficiently, and thus cope with the problem of external interference factors. We performed experiments on the Dataset of Object deTection in Aerial images (DOTA) dataset and the HRSC2016 dataset; the accuracy reached 71.5% mAP, which exceeds most object detectors of one-stage and two-stages at present. Meanwhile, the size of our model is only 9.2 M, which satisfies the requirement of being lightweight. The experimental results demonstrate that our method not only has better detection accuracy but also maintains high efficiency in RSIs. |
| first_indexed | 2024-03-09T19:44:35Z |
| format | Article |
| id | doaj.art-9a6c6c69b4964cd4816f90c39d7c62e9 |
| institution | Directory Open Access Journal |
| issn | 2220-9964 |
| language | English |
| last_indexed | 2024-03-09T19:44:35Z |
| publishDate | 2022-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | ISPRS International Journal of Geo-Information |
| spelling | doaj.art-9a6c6c69b4964cd4816f90c39d7c62e92023-11-24T01:28:33ZengMDPI AGISPRS International Journal of Geo-Information2220-99642022-03-0111318910.3390/ijgi11030189A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation NetworkLiming Zhou0Xiaohan Rao1Yahui Li2Xianyu Zuo3Baojun Qiao4Yinghao Lin5Henan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng 475000, ChinaIn recent years, significant progress has been obtained in object detection using Convolutional Neural Networks (CNNs). However, owing to the particularity of Remote Sensing Images (RSIs), common object detection methods are not well suited for RSIs. Aiming at the difficulties in RSIs, this paper proposes an object detection method based on the Dense Feature Fusion Path Aggregation Network (DFF-PANet). Firstly, for better improving the detection performance of small and medium-sized instances, we propose Feature Reuse Module (FRM), which can integrate semantic and location information contained in feature maps; this module can reuse feature maps in the backbone to enhance the detection capability of small and medium-sized instances. After that, we design the DFF-PANet, which can help feature information extracted from the backbone to be fused more efficiently, and thus cope with the problem of external interference factors. We performed experiments on the Dataset of Object deTection in Aerial images (DOTA) dataset and the HRSC2016 dataset; the accuracy reached 71.5% mAP, which exceeds most object detectors of one-stage and two-stages at present. Meanwhile, the size of our model is only 9.2 M, which satisfies the requirement of being lightweight. The experimental results demonstrate that our method not only has better detection accuracy but also maintains high efficiency in RSIs.https://www.mdpi.com/2220-9964/11/3/189feature reuse moduleresidual dense blockdense feature fusionremote sensing |
| spellingShingle | Liming Zhou Xiaohan Rao Yahui Li Xianyu Zuo Baojun Qiao Yinghao Lin A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network ISPRS International Journal of Geo-Information feature reuse module residual dense block dense feature fusion remote sensing |
| title | A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network |
| title_full | A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network |
| title_fullStr | A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network |
| title_full_unstemmed | A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network |
| title_short | A Lightweight Object Detection Method in Aerial Images Based on Dense Feature Fusion Path Aggregation Network |
| title_sort | lightweight object detection method in aerial images based on dense feature fusion path aggregation network |
| topic | feature reuse module residual dense block dense feature fusion remote sensing |
| url | https://www.mdpi.com/2220-9964/11/3/189 |
| work_keys_str_mv | AT limingzhou alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT xiaohanrao alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT yahuili alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT xianyuzuo alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT baojunqiao alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT yinghaolin alightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT limingzhou lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT xiaohanrao lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT yahuili lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT xianyuzuo lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT baojunqiao lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork AT yinghaolin lightweightobjectdetectionmethodinaerialimagesbasedondensefeaturefusionpathaggregationnetwork |