Next Article in Journal
Assessing the Reliability of Satellite and Reanalysis Estimates of Rainfall in Equatorial Africa
Next Article in Special Issue
Multi-Object Segmentation in Complex Urban Scenes from High-Resolution Remote Sensing Data
Previous Article in Journal
Spatiotemporal Changes and Driving Factors of Cultivated Soil Organic Carbon in Northern China’s Typical Agro-Pastoral Ecotone in the Last 30 Years
Previous Article in Special Issue
Making Low-Resolution Satellite Images Reborn: A Deep Learning Approach for Super-Resolution Building Extraction
Article

Learning Adjustable Reduced Downsampling Network for Small Object Detection in Urban Environments

1
Department of Geography, University of California, Santa Barbara, CA 93106, USA
2
Department of Geography, San Diego State University, San Diego, CA 92182, USA
3
Center for Complex Human-Environment Systems, San Diego State University, San Diego, CA 92182, USA
4
Department of Computer Science, San Diego State University, San Diego, CA 92182, USA
*
Author to whom correspondence should be addressed.
Academic Editors: Weijia Li, Lichao Mou, Angelica I. Aviles-Rivero, Runmin Dong and Juepeng Zheng
Remote Sens. 2021, 13(18), 3608; https://0-doi-org.brum.beds.ac.uk/10.3390/rs13183608
Received: 10 August 2021 / Revised: 5 September 2021 / Accepted: 7 September 2021 / Published: 10 September 2021
(This article belongs to the Special Issue Deep Learning in Remote Sensing Application)
Detecting small objects (e.g., manhole covers, license plates, and roadside milestones) in urban images is a long-standing challenge mainly due to the scale of small object and background clutter. Although convolution neural network (CNN)-based methods have made significant progress and achieved impressive results in generic object detection, the problem of small object detection remains unsolved. To address this challenge, in this study we developed an end-to-end network architecture that has three significant characteristics compared to previous works. First, we designed a backbone network module, namely Reduced Downsampling Network (RD-Net), to extract informative feature representations with high spatial resolutions and preserve local information for small objects. Second, we introduced an Adjustable Sample Selection (ADSS) module which frees the Intersection-over-Union (IoU) threshold hyperparameters and defines positive and negative training samples based on statistical characteristics between generated anchors and ground reference bounding boxes. Third, we incorporated the generalized Intersection-over-Union (GIoU) loss for bounding box regression, which efficiently bridges the gap between distance-based optimization loss and area-based evaluation metrics. We demonstrated the effectiveness of our method by performing extensive experiments on the public Urban Element Detection (UED) dataset acquired by Mobile Mapping Systems (MMS). The Average Precision (AP) of the proposed method was 81.71%, representing an improvement of 1.2% compared with the popular detection framework Faster R-CNN. View Full-Text
Keywords: mobile mapping; deep learning; convolution neural network (CNN); object detection; small urban elements; reduced downsampling network; adjustable sample selection mobile mapping; deep learning; convolution neural network (CNN); object detection; small urban elements; reduced downsampling network; adjustable sample selection
Show Figures

Graphical abstract

MDPI and ACS Style

Zhang, H.; An, L.; Chu, V.W.; Stow, D.A.; Liu, X.; Ding, Q. Learning Adjustable Reduced Downsampling Network for Small Object Detection in Urban Environments. Remote Sens. 2021, 13, 3608. https://0-doi-org.brum.beds.ac.uk/10.3390/rs13183608

AMA Style

Zhang H, An L, Chu VW, Stow DA, Liu X, Ding Q. Learning Adjustable Reduced Downsampling Network for Small Object Detection in Urban Environments. Remote Sensing. 2021; 13(18):3608. https://0-doi-org.brum.beds.ac.uk/10.3390/rs13183608

Chicago/Turabian Style

Zhang, Huijie, Li An, Vena W. Chu, Douglas A. Stow, Xiaobai Liu, and Qinghua Ding. 2021. "Learning Adjustable Reduced Downsampling Network for Small Object Detection in Urban Environments" Remote Sensing 13, no. 18: 3608. https://0-doi-org.brum.beds.ac.uk/10.3390/rs13183608

Find Other Styles
Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Access Map by Country/Region

1
Back to TopTop