$\mathcal{R}^2$ -CNN: Fast Tiny Object Detection in Large-Scale Remote Sensing Images

Pang, Jiangmiao; Li, Cong; Xu, Zhihai; Feng, Hao

doi:10.1109/tgrs.2019.2899955

Cited by 216 publications

(104 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Areas of future research to expand this technique include testing whether there are performance improvements for detecting elephants by including the near infrared band and testing for which other species this is already a viable monitoring technique. More broadly, deep learning methods for detecting small objects can be further improved [83,84] and large training datasets containing images of wildlife from an aerial perspective should be developed. If satellite monitoring is applied at scale then developing methods to ensure standardised and occasional ground-truthing will be required to ensure image interpretation is accurate [25].…”

Section: Discussionmentioning

confidence: 99%

Using very high-resolution satellite imagery and deep learning to detect and count African elephants in heterogeneous landscapes

Duporge

Isupova

Reece

et al. 2020

Preprint

View full text Add to dashboard Cite

Satellites allow large-scale surveys to be conducted in short time periods with repeat surveys possible <24hrs. Very high-resolution satellite imagery has been successfully used to detect and count a number of wildlife species in open, homogeneous landscapes and seascapes where target animals have a strong contrast with their environment. However, no research to date has detected animals in complex heterogeneous environments or detected elephants from space using very high-resolution satellite imagery and deep learning.In this study we apply a Convolution Neural Network (CNN) model to automatically detect and count African elephants in a woodland savanna ecosystem in South Africa. We use WorldView-3 and 4 satellite data – the highest resolution satellite imagery commercially available. We train and test the model on eleven images from 2014-2019. We compare the performance accuracy of the CNN against human accuracy. Additionally, we apply the model on a coarser resolution satellite image (GeoEye-1) captured in Kenya to test if the algorithm can generalise to an elephant population outside of the training area.Our results show the CNN performs with high accuracy, comparable to human detection capabilities. The detection accuracy (i.e., F2 score) of the CNN models was 0.78 in heterogeneous areas and 0.73 in homogenous areas. This compares with the detection accuracy of the human labels with an averaged F2 score 0.77 in heterogeneous areas and 0.80 in homogenous areas. The CNN model can generalise to detect elephants in a different geographical location and from a lower resolution satellite.Our study demonstrates the feasibility of applying state-of-the-art satellite remote sensing and deep learning technologies for detecting and counting African elephants in heterogeneous landscapes. The study showcases the feasibility of using high resolution satellite imagery as a promising new wildlife surveying technique. Through creation of a customised training dataset and application of a Convolutional Neural Network, we have automated the detection of elephants in satellite imagery with as high accuracy as human detection capabilities. The success of the model to detect elephants outside of the training data site demonstrates the generalisability of the technique.

show abstract

Section: Discussionmentioning

confidence: 99%

Using very high-resolution satellite imagery and deep learning to detect and count African elephants in heterogeneous landscapes

Duporge

Isupova

Reece

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Object detection on remote sensing imagery has numerous prospects in various fields such as environmental regulation, surveillance, military [1,2], national security, traffic, forestry [3], oil and gas activity monitoring and other domains. There are many methods on detecting and locating objects from images that are captured using satellites or drones, but detection performance is not satisfactory on noisy and low-resolution images, especially when the objects are small [4]. Even on high-resolution imagery, the detection performance of small objects is lower compared to large objects [5].…”

Section: Problem Description and Motivationmentioning

confidence: 99%

“…On the other hand, there are various objects in satellite images like vehicles, small houses, small oil and gas storage tanks etc., only covering a small area [4]. The state-of-the-art detectors [8][9][10][11] show a significant performance gap between low-resolution images and their high-resolution counterparts due to a lack of input features for small objects [12].…”

Section: Problem Description and Motivationmentioning

confidence: 99%

Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

Rabbi

Ray

Schubert³

et al. 2020

Preprint

View full text Add to dashboard Cite

The detection performance of small objects in remote sensing images is not satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) shows remarkable image enhancement performance, but reconstructed images miss high-frequency edge information. Therefore, object detection performance degrades for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we apply a new edge-enhanced super-resolution GAN (EESRGAN) to improve the image quality of remote sensing images and use different detector networks in an end-to-end manner where detector loss is backpropagated into the EESRGAN to improve the detection performance. We propose an architecture with three components: ESRGAN, Edge Enhancement Network (EEN), and Detection network. We use residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we use the faster region-based convolutional network (FRCNN) (two-stage detector) and single-shot multi-box detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) and a self-assembled (oil and gas storage tank) satellite dataset show superior performance of our method compared to the standalone state-of-the-art object detectors.

show abstract

“…Inspired by the two-stage methods, RefineDet [21] can adjust the sizes of anchors and locations with the adoption of cascade regression and the application of an Anchor Refinement Module (ARM), and then filter out easy negative anchors to improve accuracy. Inspired by the great success of CNN-based object detection methods in natural images, a growing number of studies have been devoted recently to object detection in VHR optical aerial images [22,23]. Considering the arbitrary orientation, Cheng et al [24] proposed to learn a Rotation-Invariant CNN (RICNN) model based on R-CNN framework used for multi-class object detection.…”

Section: Introductionmentioning

confidence: 99%

Multi-Scale Feature Integrated Attention-Based Rotation Network for Object Detection in VHR Aerial Images

Yang

et al. 2020

Sensors

View full text Add to dashboard Cite

Accurate and robust detection of multi-class objects in very high resolution (VHR) aerial images has been playing a significant role in many real-world applications. The traditional detection methods have made remarkable progresses with horizontal bounding boxes (HBBs) due to CNNs. However, HBB detection methods still exhibit limitations including the missed detection and the redundant detection regions, especially for densely-distributed and strip-like objects. Besides, large scale variations and diverse background also bring in many challenges. Aiming to address these problems, an effective region-based object detection framework named Multi-scale Feature Integration Attention Rotation Network (MFIAR-Net) is proposed for aerial images with oriented bounding boxes (OBBs), which promotes the integration of the inherent multi-scale pyramid features to generate a discriminative feature map. Meanwhile, the double-path feature attention network supervised by the mask information of ground truth is introduced to guide the network to focus on object regions and suppress the irrelevant noise. To boost the rotation regression and classification performance, we present a robust Rotation Detection Network, which can generate efficient OBB representation. Extensive experiments and comprehensive evaluations on two publicly available datasets demonstrate the effectiveness of the proposed framework.

show abstract

$\mathcal{R}^2$ -CNN: Fast Tiny Object Detection in Large-Scale Remote Sensing Images

Cited by 216 publications

References 42 publications

Using very high-resolution satellite imagery and deep learning to detect and count African elephants in heterogeneous landscapes

Using very high-resolution satellite imagery and deep learning to detect and count African elephants in heterogeneous landscapes

Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

Multi-Scale Feature Integrated Attention-Based Rotation Network for Object Detection in VHR Aerial Images

Contact Info

Product

Resources

About