Improved YOLO-V3 with DenseNet for Multi-Scale Remote Sensing Target Detection
Open accessThe DOI matches its Crossref record; 54 of the 55 references it cites were found in a registry or an index.
Abstract
Remote sensing targets have different dimensions, and they have the characteristics of dense distribution and a complex background. This makes remote sensing target detection difficult. With the aim at detecting remote sensing targets at different scales, a new You Only Look Once (YOLO)-V3-based model was proposed. YOLO-V3 is a new version of YOLO. Aiming at the defect of poor performance of YOLO-V3 in detecting remote sensing targets, we adopted DenseNet (Densely Connected Network) to enhance feature extraction capability. Moreover, the detection scales were increased to four based on the original YOLO-V3. The experiment on RSOD (Remote Sensing Object Detection) dataset and UCS-AOD (Dataset of Object Detection in Aerial Images) dataset showed that our approach performed better than Faster-RCNN, SSD (Single Shot Multibox Detector), YOLO-V3, and YOLO-V3 tiny in terms of accuracy. Compared with original YOLO-V3, the mAP (mean Average Precision) of our approach increased from 77.10% to 88.73% in the RSOD dataset. In particular, the mAP of detecting targets like aircrafts, which are mainly made up of small targets increased by 12.12%. In addition, the detection speed was not significantly reduced. Generally speaking, our approach achieved higher accuracy and gave considerations to real-time performance simultaneously for remote sensing target detection.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures









Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472
You Only Look Once: Unified, Real-Time Object Detection
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.91 · Cited by 53,803 · cited 1×
“The other ones are one-stage target detection algorithms including SSD (Single Shot Multibox Detector) [31–33], DSSD (Deconvolution Single Shot Multibox Detector) [34], FSSD (Feature Fusion Single Shot Multibox Detector) [35], YOLO [36], YOLO-V2 [37], and YOLO-V3 [38].”
BackgroundRich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684
(2016) · Cited by 31,590
He, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.322 · Cited by 29,137
Girshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
SSD: Single Shot MultiBox Detector
Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.-Y., & Berg, A. C. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46448-0_2 · Cited by 22,128
YOLOv3: An Incremental Improvement
Redmon, J., & Farhadi, A. (2018) · arXiv · doi:10.48550/arxiv.1804.02767 · Cited by 17,141
AugFPN: Improving Multi-Scale Feature Learning for Object Detection
Guo, C., Fan, B., Zhang, Q., Xiang, S., & Pan, C. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.01261 · Cited by 588 · cited 3×
“Abbreviations: The abbreviations in this paper are as follows: YOLO You Only Look Once CV Computer Version SVM Support Vector Machine HOG Histograms of Oriented Gradients DPM Deformable Parts Model IOU Intersection over Union FC Full Connected Layer FCN Full Convolutional Network CNN Convolutional Neural Network GT Gr…”
MethodYOLO v3-Tiny: Object Detection and Recognition using one stage improved model
Adarsh, P., Rathi, P., & Kumar, M. (2020) · 2020 6th International Conference on Advanced Computing and Communication Systems (ICACCS) · doi:10.1109/icaccs48705.2020.9074315 · Cited by 545 · cited 1×
“At present, numerous state-of-the-art target detection models with higher speed are proposed based on YOLO such as YOLO-V3 tiny [39] and TF-YOLO [40].”
BackgroundYOLO-Tomato: A Robust Algorithm for Tomato Detection Based on YOLOv3
Liu, G., Nouaze, J. C., Touko Mbouembe, P. L., & Kim, J. H. (2020) · Sensors · doi:10.3390/s20072145 · Cited by 417 · cited 1×
“Reference [51] adopted circular ground truth to realize tomato detection.”
MethodDC-SPP-YOLO: Dense connection and spatial pyramid pooling based YOLO for object detection
Huang, Z., Wang, J., Fu, X., Yu, T., Guo, Y., & Wang, R. (2020) · Information Sciences · doi:10.1016/j.ins.2020.02.067 · Cited by 379
UAV-YOLO: Small Object Detection on Unmanned Aerial Vehicle Perspective
Liu, M., Wang, X., Zhou, A., Fu, X., Ma, Y., & Piao, C. (2020) · Sensors · doi:10.3390/s20082238 · Cited by 363 · cited 6×
“3 UAV-YOLO [52] Figure 1 in Reference [52] 75.”
BackgroundInfluentialZhu, Y., & Newsam, S. (2017) · 2017 IEEE International Conference on Image Processing (ICIP) · doi:10.1109/icip.2017.8296389 · Cited by 314 · cited 2×
“Unlike ResNet, which adds the values of the s bsequent layers by co structing an identity map, DenseNet [53] connects all the layers for chan el merging to achieve feature reuse.”
Automatic License Plate Recognition via sliding-window darknet-YOLO deep learning
Hendry, & Chen, R.-C. (2019) · Image and Vision Computing · doi:10.1016/j.imavis.2019.04.007 · Cited by 304
Wang, X., Chen, H., Chen, Y., Ma, M., Zhang, K., Li, F., Zheng, Y., Zeng, D., Wang, Q., & Shi, J. (2012) · Advanced Materials · doi:10.1002/adma.201104033 · Cited by 239
Hashimoto, R., Requa, J., Dao, T., Ninh, A., Tran, E., Mai, D., Lugo, M., El-Hage Chehade, N., Chang, K. J., Karnes, W. E., et al. (2020) · Gastrointestinal Endoscopy · doi:10.1016/j.gie.2019.12.049 · Cited by 236
Sun, H., Sun, X., Wang, H., Li, Y., & Li, X. (2012) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2011.2161569 · Cited by 219 · cited 1×
“[43] extracted the region of interest with sliding windows, and then used the features of Bag-of-Words to detect the targets.”
MethodXiao, B., Xu, Y., Bi, X., Zhang, J., & Ma, X. (2020) · Neurocomputing · doi:10.1016/j.neucom.2018.09.101 · Cited by 174
Application Research of Improved YOLO V3 Algorithm in PCB Electronic Component Detection
Li, J., Gu, J., Huang, Z., & Wen, J. (2019) · Applied Sciences · doi:10.3390/app9183750 · Cited by 162 · cited 1×
“References [49,50] adopted four detection layers to enhance the performance of detecting small targets.”
MethodThe Application of Improved YOLO V3 in Multi-Scale Target Detection
Ju, M., Luo, H., Wang, Z., Hui, B., & Chang, Z. (2019) · Applied Sciences · doi:10.3390/app9183775 · Cited by 136 · cited 1×
“References [49,50] adopted four detection layers to enhance the performance of detecting small targets.”
MethodMDSSD: multi-scale deconvolutional single shot detector for small objects
Cui, L., Ma, R., Lv, P., Jiang, X., Gao, Z., Zhou, B., & Xu, M. (2020) · Science China Information Sciences · doi:10.1007/s11432-019-2723-1 · Cited by 123
Mask R-CNN With Pyramid Attention Network for Scene Text Detection
Huang, Z., Zhong, Z., Sun, L., & Huo, Q. (2019) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2019.00086 · Cited by 114
Airport Detection on Optical Satellite Images Using Deep Convolutional Neural Networks
Zhang, P., Niu, X., Dou, Y., & Xia, F. (2017) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2017.2673118 · Cited by 86
TF-YOLO: An Improved Incremental Network for Real-Time Object Detection
He, W., Huang, Z., Wei, Z., Li, C., & Guo, B. (2019) · Applied Sciences · doi:10.3390/app9163225 · Cited by 79 · cited 2×
“Reference [40] simplified the feature extraction network to obtain faster detection speed and adopted multiple layers concatenation to enhance the performance of feature extraction.”
BackgroundMethodKujawa, S., Mazurkiewicz, J., & Czekała, W. (2020) · Journal of Cleaner Production · doi:10.1016/j.jclepro.2020.120814 · Cited by 77
Faster R-CNN Based Table Detection Combining Corner Locating
Sun, N., Zhu, Y., & Hu, X. (2019) · IAPR International Conference on Document Analysis and Recognition (ICDAR) · doi:10.1109/icdar.2019.00212 · Cited by 58 · cited 5×
“More commonly, researchers use existing target detection algorithms such as Faster R-CNN in remote sensing target detection tasks.”
MethodInfluentialOracle Bone Inscription Detector Based on SSD
Meng, L., Lyu, B., Zhang, Z., Aravinda, C. V., Kamitoku, N., & Yamazaki, K. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-30754-7_13 · Cited by 57
Bilal, M., & Hanif, M. S. (2019) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2019.2906132 · Cited by 45 · cited 1×
“Compared with traditional target detection algorithms like HOG-SVM (Histogram of Oriented Gradients-Support Vector Machine) [12,13], DPM (Deformable Parts Model) [14,15], and HOG-Cascade [16,17], CNN-based target detection algorithms have great advantages in many aspects such as speed and accuracy.”
MethodRemote Sensing Airport Detection Based on End-to-End Deep Transferable Convolutional Neural Networks
Li, S., Xu, Y., Zhu, M., Ma, S., & Tang, H. (2019) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2019.2904076 · Cited by 36
Research on Vehicle Parts Defect Detection Based on Deep Learning
Liqun, W., Jiansheng, W., & Dingjin, W. (2020) · Journal of Physics Conference Series · doi:10.1088/1742-6596/1437/1/012004 · Cited by 32 · cited 1×
“Compared with traditional target detection algorithms like HOG-SVM (Histogram of Oriented Gradients-Support Vector Machine) [12,13], DPM (Deformable Parts Model) [14,15], and HOG-Cascade [16,17], CNN-based target detection algorithms have great advantages in many aspects such as speed and accuracy.”
MethodA Multivariate Hit-or-Miss Transform for Conjoint Spatial and Spectral Template Matching
Weber, J., & Lefèvre, S. (2008) · Lecture notes in computer science · doi:10.1007/978-3-540-69905-7_26 · Cited by 31 · cited 1×
“[41] proposed a method of making use of image analysis to extract coastline templates and adopted this method to detect oil tanks.”
MethodZeng, Y., Ritz, C. H., Zhao, J., & Lan, J. (2020) · Remote Sensing · doi:10.3390/rs12030400 · Cited by 30
Dim small target detection based on convolutinal neural network in star image
Xue, D., Sun, J., Hu, Y., Zheng, Y., Zhu, Y., & Zhang, Y. (2019) · Multimedia Tools and Applications · doi:10.1007/s11042-019-7412-z · Cited by 27
Zhu, H., Zhang, P., Wang, L., Zhang, X., & Jiao, L. (2019) · Remote Sensing Letters · doi:10.1080/2150704x.2019.1633486 · Cited by 27
Application Research of YOLO v2 Combined with Color Identification
Zhang, X., Qiu, Z., Huang, P., Hu, J., & Luo, J. (2018) · International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC) · doi:10.1109/cyberc.2018.00036 · Cited by 26 · cited 8×
“YOLO-V3 is a further improved version based on YOLO-V2 by upgrading the original Darknet19 to Darknet53 and adopts multi-scale detection layers (three scales) to detect the targets.”
BackgroundInfluentialRotation-reversal invariant HOG cascade for facial expression recognition
Chen, J., Takiguchi, T., & Ariki, Y. (2017) · Signal Image and Video Processing · doi:10.1007/s11760-017-1111-x · Cited by 19
Automated Ship Detection from Optical Remote Sensing Images
Yu, Y., Yang, X., Xiao, S., & Lin, J. (2012) · Key engineering materials · doi:10.4028/www.scientific.net/kem.500.785 · Cited by 18 · cited 1×
“[45] combined the prior characteristics of the airports and coasts, respectively, with deep learning to conduct remote sensing target detection.”
BackgroundVehicle target detection methods based on color fusion deformable part model
Zhang, D. (2018) · EURASIP Journal on Wireless Communications and Networking · doi:10.1186/s13638-018-1111-8 · Cited by 15 · cited 2×
“Abbreviations: The abbreviations in this paper are as follows: YOLO You Only Look Once CV Computer Version SVM Support Vector Machine HOG Histograms of Oriented Gradients DPM Deformable Parts Model IOU Intersection over Union FC Full Connected Layer FCN Full Convolutional Network CNN Convolutional Neural Network GT Gr…”
MethodA two-stage framework of target detection in high-resolution hyperspectral images
Li, Y., Xu, J., Xia, R., Wang, X., & Xie, W. (2019) · Signal Image and Video Processing · doi:10.1007/s11760-019-01470-z · Cited by 12
Second-Generation Sequencing with Deep Reinforcement Learning for Lung Infection Detection
Liu, Z., Zhang, G., Jingyuan, Z., Yu, L., Sheng, J., Zhang, N., & Yuan, H. (2020) · Journal of Healthcare Engineering · doi:10.1155/2020/3264801 · Cited by 12
Shih, K.-H., Chiu, C., & Pu, Y.-Y. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8683842 · Cited by 10
Multi-feature Based Automated Flower Harvesting Techniques in Deep Convolutional Neural Networking
Shree, C., Kaur, R. P., Upadhyay, S., & Joshi, J. (2019) · International Conference on Internet of Things: Smart Innovation and Usages (IoT-SIU) · doi:10.1109/iot-siu.2019.8777338 · Cited by 10
Adaptive learning feature pyramid for object detection
Wong, F., & Hu, H. (2019) · IET Computer Vision · doi:10.1049/iet-cvi.2018.5654 · Cited by 10 · cited 3×
“Abbreviations: The abbreviations in this paper are as follows: YOLO You Only Look Once CV Computer Version SVM Support Vector Machine HOG Histograms of Oriented Gradients DPM Deformable Parts Model IOU Intersection over Union FC Full Connected Layer FCN Full Convolutional Network CNN Convolutional Neural Network GT Gr…”
MethodFeng, T., Ma, H., Cheng, X., & Zhang, H. (2018) · Journal of Applied Remote Sensing · doi:10.1117/1.jrs.12.025006 · Cited by 7 · cited 1×
“[42] proposed the algorithm named multi-resolution segmentation, which segmented remote sensing images into multiple regions for detection by three parameters: shape, scale, and density.”
MethodRPN-FCN based Rust detection on power equipment
Yuan, J., Xue, B., Zhang, W., Xu, L., Sun, H., & Zhou, J. (2019) · Procedia Computer Science · doi:10.1016/j.procs.2019.01.236 · Cited by 7
Position-Free Hand Gesture Recognition Using Single Shot MultiBox Detector Based Neural Network
Tang, J., Yao, X., Kang, X., Nishide, S., & Ren, F. (2019) · IEEE International Conference on Mechatronics and Automation (ICMA) · doi:10.1109/icma.2019.8816203 · Cited by 7
Xu, Z., Huo, Y., Liu, K., & Liu, S. (2020) · International Journal of Distributed Sensor Networks · doi:10.1177/1550147720912959 · Cited by 7
Shi, W., Jiang, J., Bao, S., & Tan, D. (2019) · Applied Sciences · doi:10.3390/app9224836 · Cited by 6
Real-time pedestrian detection based on GMM and HOG cascade
Jin, M., Jeong, K., Yoon, S., & Park, D. S. (2013) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2051382 · Cited by 5 · cited 1×
“Compared with traditional target detection algorithms like HOG-SVM (Histogram of Oriented Gradients-Support Vector Machine) [12,13], DPM (Deformable Parts Model) [14,15], and HOG-Cascade [16,17], CNN-based target detection algorithms have great advantages in many aspects such as speed and accuracy.”
MethodShi, Y., Wang, W., Gong, Q., & Li, D. (2019) · The Journal of Engineering · doi:10.1049/joe.2019.0240 · Cited by 5
Haque, M. F., & Kang, D.-S. (2018) · The Journal of Korean Institute of Information Technology · doi:10.14801/jkiit.2018.16.10.93 · Cited by 4 · cited 1×
“The other ones are one-stage target detection algorithms including SSD (Single Shot Multibox Detector) [31–33], DSSD (Deconvolution Single Shot Multibox Detector) [34], FSSD (Feature Fusion Single Shot Multibox Detector) [35], YOLO [36], YOLO-V2 [37], and YOLO-V3 [38].”
BackgroundCollaborative Cross-Domain $k$ NN Search for Remote Sensing Image Processing
Zhong, Y. C., Weng, W. H., Li, J., & Zhu, S. (2019) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2019.2906686 · Cited by 4
SLIC segmentation method for full‐polarised remote‐sensing image
Zhang, Z., Chen, J., & Liu, Z. (2019) · The Journal of Engineering · doi:10.1049/joe.2019.0337 · Cited by 4
Research works printed but not linked: 4
- Shen, 2017, Building Detection from High Resolution Remote Sensing Imagery Based on a Deformable Part Model, Geomat. Inf. Sci. Wuhan Univ., 42, 1285
- Li, X., Shang, M., Qin, H., and Chen, L. (2015). Fast Accurate Fish Detection and Recognition of Underwater Images with Fast R-CNN, IEEE.
- Kaiming, H., Gkioxari, G., Dollar, P., and Girshick, R. (2017, January 22–29). Mask R-CNN. Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy.
- Bochkovskiy, A., Chien-Yao, W., and Liao, H.Y.M. (2020). YOLOv4: Optimal speed and accuracy of object detection. arXiv.
Cited by
Comparative analysis of deep learning image detection algorithms
Srivastava, S., Divekar, A. V., Anilkumar, C., Naik, I., Kulkarni, V., & Pattabiraman, V. (2021) · Journal of Big Data · doi:10.1186/s40537-021-00434-w · Cited by 350
YOLO-SE: Improved YOLOv8 for Remote Sensing Object Detection and Recognition
Wu, T., & Dong, Y. (2023) · Applied Sciences · doi:10.3390/app132412977 · Cited by 210
“Xu et al. [22] proposed a multi-scale remote sensing object detection model based on YOLOv3.”
MethodA Review of Practical AI for Remote Sensing in Earth Sciences
Janga, B., Asamani, G., Sun, Z., & Cristea, N. (2023) · Remote Sensing · doi:10.3390/rs15164112 · Cited by 194
“YOLOv2 [100] improves the speed and the type of object detected, and YOLOv3 enables the prediction of objects of different sizes [101–103].”
BackgroundYOLO-Extract: Improved YOLOv5 for Aircraft Object Detection in Remote Sensing Images
Liu, Z., Gao, Y., Du, Q., Chen, M., & Lv, W. (2023) · IEEE Access · doi:10.1109/access.2023.3233964 · Cited by 135
“Reference [9] proposed to use DenseNet [10] to improve YOLOv3 [7] and improve the detection accuracy of remote sensing images by improving the structure in Backbone, but the structure of DenseNet is too complex and the number of parameters is too large, which leads to a decrease in detection speed.”
MethodLiu, C., Wu, Y., Liu, J., & Sun, Z. (2021) · Electronics · doi:10.3390/electronics10070771 · Cited by 114
Self-citation
Metadata
- Authors
- Xu, D., & Wu, Y.
- Affiliations
- Nanjing University of Aeronautics and Astronautics (CN)
- Corresponding author
- Wu, Yiquan
- ORCID
- Wu, Yiquan
- Venue
- Sensors
- Publisher
- MDPI AG
- Published
- 2020-07-31
- Volume
- 20
- Issue
- 15
- Pages
- 4276
- ISSN
- 1424-8220 (ISSN-L)
- PMC
- PMC7435986
- OpenAlex
- W3046857037
- Cited by
- 141
- References
- 55
- Citations by year
- 2026: 5 · 2025: 21 · 2024: 27 · 2023: 32 · 2022: 32 · 2021: 20 · 2020: 4
- FWCI
- 8.55
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Sensors (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- National Natural Science Foundation of China (61573183)
- National Laboratory of Pattern Recognition (201900029)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- YOLOv3, DenseNet, multi-scale object detection, feature extraction, aerial images, small object detection, aircraft detection, real-time detection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.