Moving object detection in video satellite image based on deep learning

Zhang, Xueyang; Xiang, Junhua

doi:10.1117/12.2296714

Cited by 12 publications

(10 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent literature [22,23,35,15,31,8,27,17,34,13,2,29,30] also suggests to apply deep learning on aerial and satellite high resolution RGB single images, however, the work shows moderate performance for GSD larger than 15 cm [23]. All work is also tested with rather narrow datasets of very different sceneries which makes the validity of the results questionable and the comparison of methods difficult.…”

Section: Introductionmentioning

confidence: 99%

On Learning Vehicle Detection in Satellite Video

Pflugfelder,

Weissenfeld,

Wagner

2020

Preprint

View full text Add to dashboard Cite

Vehicle detection in aerial and satellite images is still challenging due to their tiny appearance in pixels compared to the overall size of remote sensing imagery. Classical methods of object detection very often fail in this scenario due to violation of implicit assumptions made such as rich texture, small to moderate ratios between image size and object size. Satellite video is a very new modality which introduces temporal consistency as inductive bias. Approaches for vehicle detection in satellite video use either background subtraction, frame differencing or subspace methods showing moderate performance (0.26 -0.82 F 1 score). This work proposes to apply recent work on deep learning for wide-area motion imagery (WAMI) on satellite video. We show in a first approach comparable results (0.84 F 1 ) on Planet's SkySat-1 LasVegas video with room for further improvement.

show abstract

Section: Introductionmentioning

confidence: 99%

On Learning Vehicle Detection in Satellite Video

Pflugfelder,

Weissenfeld,

Wagner

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Background subtraction and consecutive frame differencing are main approaches to detect temporal differences, but have their challenges. Background subtraction-based algorithms [1,2,3,4,5] use registered video frames to build a background model. From each frame, the algorithm subtracts the modeled background to detect changes and hence finds objects.…”

Section: Related Workmentioning

confidence: 99%

“…Recent improvements in deep neural networks and hardware capabilities have attracted more attention to these fields. Object detection and tracking applications on Wide Area Motion Imagery (WAMI) have had its share of the attention [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16], and today it still holds its place as a challenging computer vision task. WAMI deals with monitoring a large area, which is several kilometers in diameter, with an airborne sensor consisting of multiple cameras.…”

Section: Introductionmentioning

confidence: 99%

HM-Net: A Regression Network for Object Center Detection and Tracking on Wide Area Motion Imagery

Motorcu¹,

Ates²,

Uğurdağ³

et al. 2021

Preprint

View full text Add to dashboard Cite

“…Currently, part of the satellites is equipped with high resolution cameras, and satellite surveillance using cameras is possible. In 2017, Zhang et al [4] proposed a moving object detection method on satellite RGB image. In our methods, in addition to object detection, we further propose methods for object tracking and abnormal behavior detection on these objects.…”

Section: Related Workmentioning

confidence: 99%

Abnormal Maritime Activity Detection in Satellite Image Sequences Using Trajectory Features

Fahn¹,

Ling²,

Yeh³

et al. 2019

IJFCC

View full text Add to dashboard Cite

Maritime security has gained much attention these days because of the frequent interaction between countries with marine borders. Surveillance systems on land are installed on a fixed point, while there is no mounting point for marine surveillance. Fortunately, with the advancement of aerospace technology, the maritime monitoring is possible with continuous satellite image sequences. In this paper, we proposed a complete scheme for maritime surveillance using image sequences on embedded satellite systems, including maritime object detection, tracking, and abnormal activity classification. The recall rate for object detection is 78%. The recall rate for tracking is 91%. The recall rate for anomaly detection is 78%. The FPS is one frame per second on a low power-consuming device installable on a satellite system. The proposed scheme is helpful for the national defense and is labor saving for coast guard affairs.

show abstract

Moving object detection in video satellite image based on deep learning

Cited by 12 publications

References 0 publications

On Learning Vehicle Detection in Satellite Video

On Learning Vehicle Detection in Satellite Video

HM-Net: A Regression Network for Object Center Detection and Tracking on Wide Area Motion Imagery

Abnormal Maritime Activity Detection in Satellite Image Sequences Using Trajectory Features

Contact Info

Product

Resources

About