A multiscale object detection approach for remote sensing images based on MSE-DenseNet and the dynamic anchor assignment

Zhu, Huming; Zhang, Peng; Wang, Libing; Zhang, Xiaohua; Jiao, Licheng

doi:10.1080/2150704x.2019.1633486

Cited by 23 publications

(14 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Freed full-angle replay technology brings fans a stronger visual experience, and the application of big data can improve the quality of training, measure the value of players, and ensure the fairness of the game, among others. However, it is ultimately an auxiliary tool that ultimately relies on referees to make decisions, so it is important to combine big data with experience to make the most reasonable calls [8]. Shan et al mentioned that big data can be used to collect and analyze players' height, bouncing, sprinting explosive power, and so on during the draft, to evaluate players' comprehensive ability and predict their ranking, and the fact is that most of the predicted high ranking players will also become successful.…”

Section: Related Workmentioning

confidence: 99%

[Retracted] Real‐Time Analysis of Basketball Sports Data Based on Deep Learning

Peng

2021

Complexity

View full text Add to dashboard Cite

This paper focuses on the theme of the application of deep learning in the field of basketball sports, using research methods such as literature research, video analysis, comparative research, and mathematical statistics to explore deep learning in real-time analysis of basketball sports data. The basketball posture action recognition and analysis system proposed for basketball movement is composed of two parts serially. The first part is based on the bottom-up posture estimation method to locate the joint points and is used to extract the posture sequence of the target in the video. The second part is the analysis and research of the action recognition algorithm based on the convolution of the space-time graph. According to the extracted posture sequence, the basketball action of the set classification is recognized. In order to obtain more accurate and three-dimensional information, a multitraining target method can be used in training; that is, multiple indicators can be detected and feedback is provided at the same time to correct player errors in time; the other is an auxiliary method, which is compared with ordinary training. The method can actively correct technical movements, train players to form muscle memory, and improve their abilities. Through the research of this article, it provides a theoretical basis for promoting the application of deep learning in the field of basketball and also provides a theoretical reference for the wider application of deep learning in the field of sports. At the same time, the designed real-time analysis system of basketball data also provides more actual reference values for coaches and athletes.

show abstract

Section: Related Workmentioning

confidence: 99%

[Retracted] Real‐Time Analysis of Basketball Sports Data Based on Deep Learning

Peng

2021

Complexity

View full text Add to dashboard Cite

show abstract

“…Second, the nine anchors of our algorithms are obtained through k-means clustering on the ground-truth bounding boxes of the training data, and the proportion of small objects is higher than the one of large objects in the training set, which causes the anchor priors to favor small size. In addition to the AP and MAP values, we calculated the average IoU of the StAN model and compared it with the algorithms in [73]. As shown in Table 3, the proposed StAN model achieves a higher location accuracy, which further proves the correctness of the detection task on the shallow layer features.…”

Section: Resultsmentioning

confidence: 86%

“…As for the DOTA dataset, only several deep algorithms are selected for comparison, including SSD [27], YOLO v2 [67], R-FCN [68], YOLO v3 [44], R-DFPN [69], RRPN [61,70] and RoI Transformer [71], ICN [72]. In addition, the result of YOLO v3 [44] and results from [73] are selected for comparison, to verify the performance of location accuracy of the proposed algorithm.…”

Section: Baseline Methodsmentioning

confidence: 99%

Subtask Attention Based Object Detection in Remote Sensing Images

Xiong

Tan

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Object detection in remote sensing images (RSIs) is one of the basic tasks in the field of remote sensing image automatic interpretation. In recent years, the deep object detection frameworks of natural scene images (NSIs) have been introduced into object detection on RSIs, and the detection performance has improved significantly because of the powerful feature representation. However, there are still many challenges concerning the particularities of remote sensing objects. One of the main challenges is the missed detection of small objects which have less than five percent of the pixels of the big objects. Generally, the existing algorithms choose to deal with this problem by multi-scale feature fusion based on a feature pyramid. However, the benefits of this strategy are limited, considering that the location of small objects in the feature map will disappear when the detection task is processed at the end of the network. In this study, we propose a subtask attention network (StAN), which handles the detection task directly on the shallow layer of the network. First, StAN contains one shared feature branch and two subtask attention branches of a semantic auxiliary subtask and a detection subtask based on the multi-task attention network (MTAN). Second, the detection branch uses only low-level features considering small objects. Third, the attention map guidance mechanism is put forward to optimize the network for keeping the identification ability. Fourth, the multi-dimensional sampling module (MdS), global multi-view channel weights (GMulW) and target-guided pixel attention (TPA) are designed for further improvement of the detection accuracy in complex scenes. The experimental results on the NWPU VHR-10 dataset and DOTA dataset demonstrated that the proposed algorithm achieved the SOTA performance, and the missed detection of small objects decreased. On the other hand, ablation experiments also proved the effects of MdS, GMulW and TPA.

show abstract

“…Recently, remote sensing images [1][2][3][4] have attracted more research in the field of computer version (CV) with the rapid development of satellite and imaging technology. There is a significant value on information extraction of remote sensing images.…”

Section: Introductionmentioning

confidence: 99%

Improved YOLO-V3 with DenseNet for Multi-Scale Remote Sensing Target Detection

2020

Sensors

View full text Add to dashboard Cite

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.

show abstract

A multiscale object detection approach for remote sensing images based on MSE-DenseNet and the dynamic anchor assignment

Cited by 23 publications

References 8 publications

[Retracted] Real‐Time Analysis of Basketball Sports Data Based on Deep Learning

[Retracted] Real‐Time Analysis of Basketball Sports Data Based on Deep Learning

Subtask Attention Based Object Detection in Remote Sensing Images

Improved YOLO-V3 with DenseNet for Multi-Scale Remote Sensing Target Detection

Contact Info

Product

Resources

About