The Experts below are selected from a list of 40149 Experts worldwide ranked by ideXlab platform
Junwei Han - One of the best experts on this subject based on the ideXlab platform.
-
object detection in Optical Remote Sensing images a survey and a new benchmark
Isprs Journal of Photogrammetry and Remote Sensing, 2020Co-Authors: Gang Wan, Gong Yi-cheng, Liqiu Meng, Junwei HanAbstract:Abstract Substantial efforts have been devoted more recently to presenting various methods for object detection in Optical Remote Sensing images. However, the current survey of datasets and deep learning based methods for object detection in Optical Remote Sensing images is not adequate. Moreover, most of the existing datasets have some shortcomings, for example, the numbers of images and object categories are small scale, and the image diversity and variations are insufficient. These limitations greatly affect the development of deep learning based object detection methods. In the paper, we provide a comprehensive review of the recent deep learning based object detection progress in both the computer vision and earth observation communities. Then, we propose a large-scale, publicly available benchmark for object DetectIon in Optical Remote Sensing images, which we name as DIOR. The dataset contains 23,463 images and 192,472 instances, covering 20 object classes. The proposed DIOR dataset (1) is large-scale on the object categories, on the object instance number, and on the total image number; (2) has a large range of object size variations, not only in terms of spatial resolutions, but also in the aspect of inter- and intra-class size variability across objects; (3) holds big variations as the images are obtained with different imaging conditions, weathers, seasons, and image quality; and (4) has high inter-class similarity and intra-class diversity. The proposed benchmark can help the researchers to develop and validate their data-driven methods. Finally, we evaluate several state-of-the-art approaches on our DIOR dataset to establish a baseline for future research.
-
learning rotation invariant convolutional neural networks for object detection in vhr Optical Remote Sensing images
IEEE Transactions on Geoscience and Remote Sensing, 2016Co-Authors: Gong Yi-cheng, Peicheng Zhou, Junwei HanAbstract:Object detection in very high resolution Optical Remote Sensing images is a fundamental problem faced for Remote Sensing image analysis. Due to the advances of powerful feature representations, machine-learning-based object detection is receiving increasing attention. Although numerous feature representations exist, most of them are handcrafted or shallow-learning-based features. As the object detection task becomes more challenging, their description capability becomes limited or even impoverished. More recently, deep learning algorithms, especially convolutional neural networks (CNNs), have shown their much stronger feature representation power in computer vision. Despite the progress made in nature scene images, it is problematic to directly use the CNN feature for object detection in Optical Remote Sensing images because it is difficult to effectively deal with the problem of object rotation variations. To address this problem, this paper proposes a novel and effective approach to learn a rotation-invariant CNN (RICNN) model for advancing the performance of object detection, which is achieved by introducing and learning a new rotation-invariant layer on the basis of the existing CNN architectures. However, different from the training of traditional CNN models that only optimizes the multinomial logistic regression objective, our RICNN model is trained by optimizing a new objective function via imposing a regularization constraint, which explicitly enforces the feature representations of the training samples before and after rotating to be mapped close to each other, hence achieving rotation invariance. To facilitate training, we first train the rotation-invariant layer and then domain-specifically fine-tune the whole RICNN network to further boost the performance. Comprehensive evaluations on a publicly available ten-class object detection data set demonstrate the effectiveness of the proposed method.
-
A survey on object detection in Optical Remote Sensing images
ISPRS Journal of Photogrammetry and Remote Sensing, 2016Co-Authors: Gong Yi-cheng, Junwei HanAbstract:Object detection in Optical Remote Sensing images, being a fundamental but challenging problem in the field of aerial and satellite image analysis, plays an important role for a wide range of applications and is receiving significant attention in recent years. While enormous methods exist, a deep review of the literature concerning generic object detection is still lacking. This paper aims to provide a review of the recent progress in this field. Different from several previously published surveys that focus on a specific object class such as building and road, we concentrate on more generic object categories including, but are not limited to, road, building, tree, vehicle, ship, airport, urban-area. Covering about 270 publications we survey (1) template matching-based object detection methods, (2) knowledge-based object detection methods, (3) object-based image analysis (OBIA)-based object detection methods, (4) machine learning-based object detection methods, and (5) five publicly available datasets and three standard evaluation metrics. We also discuss the challenges of current studies and propose two promising research directions, namely deep learning-based feature representation and weakly supervised learning-based geospatial object detection. It is our hope that this survey will be beneficial for the researchers to have better understanding of this research field.
-
object detection in Optical Remote Sensing images based on weakly supervised learning and high level feature learning
IEEE Transactions on Geoscience and Remote Sensing, 2015Co-Authors: Junwei Han, Gong Yi-cheng, Dingwen Zhang, Lei Guo, Jinchang RenAbstract:The abundant spatial and contextual information provided by the advanced Remote Sensing technology has facilitated subsequent automatic interpretation of the Optical Remote Sensing images (RSIs). In this paper, a novel and effective geospatial object detection framework is proposed by combining the weakly supervised learning (WSL) and high-level feature learning. First, deep Boltzmann machine is adopted to infer the spatial and structural information encoded in the low-level and middle-level features to effectively describe objects in Optical RSIs. Then, a novel WSL approach is presented to object detection where the training sets require only binary labels indicating whether an image contains the target object or not. Based on the learnt high-level features, it jointly integrates saliency, intraclass compactness, and interclass separability in a Bayesian framework to initialize a set of training examples from weakly labeled images and start iterative learning of the object detector. A novel evaluation criterion is also developed to detect model drift and cease the iterative learning. Comprehensive experiments on three Optical RSI data sets have demonstrated the efficacy of the proposed approach in benchmarking with several state-of-the-art supervised-learning-based object detection approaches.
Gong Yi-cheng - One of the best experts on this subject based on the ideXlab platform.
-
object detection in Optical Remote Sensing images a survey and a new benchmark
Isprs Journal of Photogrammetry and Remote Sensing, 2020Co-Authors: Gang Wan, Gong Yi-cheng, Liqiu Meng, Junwei HanAbstract:Abstract Substantial efforts have been devoted more recently to presenting various methods for object detection in Optical Remote Sensing images. However, the current survey of datasets and deep learning based methods for object detection in Optical Remote Sensing images is not adequate. Moreover, most of the existing datasets have some shortcomings, for example, the numbers of images and object categories are small scale, and the image diversity and variations are insufficient. These limitations greatly affect the development of deep learning based object detection methods. In the paper, we provide a comprehensive review of the recent deep learning based object detection progress in both the computer vision and earth observation communities. Then, we propose a large-scale, publicly available benchmark for object DetectIon in Optical Remote Sensing images, which we name as DIOR. The dataset contains 23,463 images and 192,472 instances, covering 20 object classes. The proposed DIOR dataset (1) is large-scale on the object categories, on the object instance number, and on the total image number; (2) has a large range of object size variations, not only in terms of spatial resolutions, but also in the aspect of inter- and intra-class size variability across objects; (3) holds big variations as the images are obtained with different imaging conditions, weathers, seasons, and image quality; and (4) has high inter-class similarity and intra-class diversity. The proposed benchmark can help the researchers to develop and validate their data-driven methods. Finally, we evaluate several state-of-the-art approaches on our DIOR dataset to establish a baseline for future research.
-
learning rotation invariant convolutional neural networks for object detection in vhr Optical Remote Sensing images
IEEE Transactions on Geoscience and Remote Sensing, 2016Co-Authors: Gong Yi-cheng, Peicheng Zhou, Junwei HanAbstract:Object detection in very high resolution Optical Remote Sensing images is a fundamental problem faced for Remote Sensing image analysis. Due to the advances of powerful feature representations, machine-learning-based object detection is receiving increasing attention. Although numerous feature representations exist, most of them are handcrafted or shallow-learning-based features. As the object detection task becomes more challenging, their description capability becomes limited or even impoverished. More recently, deep learning algorithms, especially convolutional neural networks (CNNs), have shown their much stronger feature representation power in computer vision. Despite the progress made in nature scene images, it is problematic to directly use the CNN feature for object detection in Optical Remote Sensing images because it is difficult to effectively deal with the problem of object rotation variations. To address this problem, this paper proposes a novel and effective approach to learn a rotation-invariant CNN (RICNN) model for advancing the performance of object detection, which is achieved by introducing and learning a new rotation-invariant layer on the basis of the existing CNN architectures. However, different from the training of traditional CNN models that only optimizes the multinomial logistic regression objective, our RICNN model is trained by optimizing a new objective function via imposing a regularization constraint, which explicitly enforces the feature representations of the training samples before and after rotating to be mapped close to each other, hence achieving rotation invariance. To facilitate training, we first train the rotation-invariant layer and then domain-specifically fine-tune the whole RICNN network to further boost the performance. Comprehensive evaluations on a publicly available ten-class object detection data set demonstrate the effectiveness of the proposed method.
-
A survey on object detection in Optical Remote Sensing images
ISPRS Journal of Photogrammetry and Remote Sensing, 2016Co-Authors: Gong Yi-cheng, Junwei HanAbstract:Object detection in Optical Remote Sensing images, being a fundamental but challenging problem in the field of aerial and satellite image analysis, plays an important role for a wide range of applications and is receiving significant attention in recent years. While enormous methods exist, a deep review of the literature concerning generic object detection is still lacking. This paper aims to provide a review of the recent progress in this field. Different from several previously published surveys that focus on a specific object class such as building and road, we concentrate on more generic object categories including, but are not limited to, road, building, tree, vehicle, ship, airport, urban-area. Covering about 270 publications we survey (1) template matching-based object detection methods, (2) knowledge-based object detection methods, (3) object-based image analysis (OBIA)-based object detection methods, (4) machine learning-based object detection methods, and (5) five publicly available datasets and three standard evaluation metrics. We also discuss the challenges of current studies and propose two promising research directions, namely deep learning-based feature representation and weakly supervised learning-based geospatial object detection. It is our hope that this survey will be beneficial for the researchers to have better understanding of this research field.
-
object detection in Optical Remote Sensing images based on weakly supervised learning and high level feature learning
IEEE Transactions on Geoscience and Remote Sensing, 2015Co-Authors: Junwei Han, Gong Yi-cheng, Dingwen Zhang, Lei Guo, Jinchang RenAbstract:The abundant spatial and contextual information provided by the advanced Remote Sensing technology has facilitated subsequent automatic interpretation of the Optical Remote Sensing images (RSIs). In this paper, a novel and effective geospatial object detection framework is proposed by combining the weakly supervised learning (WSL) and high-level feature learning. First, deep Boltzmann machine is adopted to infer the spatial and structural information encoded in the low-level and middle-level features to effectively describe objects in Optical RSIs. Then, a novel WSL approach is presented to object detection where the training sets require only binary labels indicating whether an image contains the target object or not. Based on the learnt high-level features, it jointly integrates saliency, intraclass compactness, and interclass separability in a Bayesian framework to initialize a set of training examples from weakly labeled images and start iterative learning of the object detector. A novel evaluation criterion is also developed to detect model drift and cease the iterative learning. Comprehensive experiments on three Optical RSI data sets have demonstrated the efficacy of the proposed approach in benchmarking with several state-of-the-art supervised-learning-based object detection approaches.
He Chen - One of the best experts on this subject based on the ideXlab platform.
-
small sample set inshore ship detection from vhr Optical Remote Sensing images based on structured sparse representation
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020Co-Authors: Yin Zhuang, He ChenAbstract:Inshore ship detection from very high resolution (VHR) Optical Remote Sensing images has been playing a critical role in various civil and military applications. However, it brings up an important challenge, which is difficult to complete effective and robust feature extraction when valid inshore ship training sample acquired is limited, and the severe imbalance problem exists of positive and negative samples. In order to tackle the abovementioned difficulties, the structured sparse representation model (SSRM) is proposed to achieve inshore ship detection in more effectively and robustly way by circumstances of the small sample set. Here, SSRM has two steps that include inshore ship region proposal (RP) and orientation prediction (OP). Related to the RP process, the error matrix embedded in SSRM not only prevents to build the high-dimension background subdictionary and imbalance problem of positive and negative samples, but also achieves an effective intraclass robustness description of inshore ships and background. For the OP stage, the low-rank constraint of common sharing atoms in SSRM can make inshore ship direction be extracted by their sparse coding. In addition, based on RP and OP guidance, the proposed comprehensive structure voting can achieve an accurate contour detection of inshore ships. Finally, several experimental results employ that Google Earth service, HRSC 2016, and DOTA datasets proved the effectiveness of the proposed method. The results show that proposed inshore ship detection method can provide approximately 83.7% Recall and 72.3% Precision by using only over 100 positive training samples, which outperforms the state of the art methods.
-
arbitrary oriented ship detection framework in Optical Remote Sensing images
IEEE Geoscience and Remote Sensing Letters, 2018Co-Authors: Wenchao Liu, He ChenAbstract:Ship detection is a challenging problem in complex Optical Remote-Sensing images. In this letter, an effective ship detection framework in Remote-Sensing images based on the convolutional neural network is proposed. The framework is designed to predict bounding box of ship with orientation angle information. Note that the angle information which is added to bounding box regression makes bounding box accurately fit into the ship region. In order to make the model adaptable to the detection of multiscale ship targets, especially small-sized ships, we design the network with feature maps from the layers of different depths. The whole detection pipeline is a single network and achieves real-time detection for a $704 \times 704$ image with the use of Titan X GPU acceleration. Through experiments, we validate the effectiveness, robustness, and accuracy of the proposed ship detection framework in complex Remote-Sensing scenes.
Ram A Hashmonay - One of the best experts on this subject based on the ideXlab platform.
-
Optical Remote Sensing to quantify fugitive particulate mass emissions from stationary short term and mobile continuous sources part i method and examples
Environmental Science & Technology, 2011Co-Authors: Ke Du, Ram A Hashmonay, Mark J. Rood, Ellsworth J. Welton, Ravi Varma, Michael R. KemmeAbstract:The emissions of particulate matter (PM) from anthropogenic sources raise public concern. A new method is described here that was developed to complete in situ rapid response measurements of PM mass emissions from fugitive dust sources by use of Optical Remote Sensing (ORS) and an anemometer. The ORS system consists of one ground-based micropulse light detection and ranging (MPL) device that was mounted on a positioner, two open path-Fourier transform infrared (OP-FTIR) spectrometers, and two open path-laser transmissometers (OP-LT). An algorithm was formulated to compute PM light extinction profiles along each of the plume’s cross sections that were determined with the MPL. Size-specific PM mass emission factors were then calculated by integrating the light extinction profiles with particle mass extinction efficiencies (determined with the OP-FTIRs/OP-LTs) and the wind’s speed and direction. This method also quantifies the spatial and temporal variability of the plume’s PM mass concentrations across each...
-
Optical Remote Sensing to quantify fugitive particulate mass emissions from stationary short-term and mobile continuous sources: part I. Method and examples.
Environmental science & technology, 2010Co-Authors: Mark J. Rood, Ram A Hashmonay, Ellsworth J. Welton, Ravi Varma, Byung J. Kim, Michael R. KemmeAbstract:The emissions of particulate matter (PM) from anthropogenic sources raise public concern. A new method is described here that was developed to complete in situ rapid response measurements of PM mass emissions from fugitive dust sources by use of Optical Remote Sensing (ORS) and an anemometer. The ORS system consists of one ground-based micropulse light detection and ranging (MPL) device that was mounted on a positioner, two open path-Fourier transform infrared (OP-FTIR) spectrometers, and two open path-laser transmissometers (OP-LT). An algorithm was formulated to compute PM light extinction profiles along each of the plume's cross sections that were determined with the MPL. Size-specific PM mass emission factors were then calculated by integrating the light extinction profiles with particle mass extinction efficiencies (determined with the OP-FTIRs/OP-LTs) and the wind's speed and direction. This method also quantifies the spatial and temporal variability of the plume's PM mass concentrations across each of the plume's cross sections. Example results from three field studies are also described to demonstrate how this new method is used to determine mass emission factors as well as characterize the dust plumes' horizontal and vertical dimensions and temporal variability of the PM's mass concentration.
-
a novel method to quantify fugitive dust emissions using Optical Remote Sensing
2008Co-Authors: Ram A Hashmonay, Mark J. Rood, Ravi Varma, Ke Du, Michael R. KemmeAbstract:Abstract : This paper describes a new method for retrieving path-averaged mass concentrations from multi-spectral light extinction measured by Optical Remote Sensing (ORS) instruments. The light extinction measurements as a function of wavelength were used in conjunction with an iterative inverse-Mie algorithm to retrieve path-averaged particulate matter (PM) mass distribution. Conventional mass concentration measurements in a controlled release experiment were used to calibrate the ORS method. A backscattering micro pulse lidar (MPL) was used to obtain the horizontal extent of the plume along MPL's line of sight. This method was used to measure concentrations and mass emission rates of PM with diameters smaller or equal to 10 microns (PM(sub 10)) and PM with diameters smaller or equal to 2.5 microns (PM(sub 2.50) that were caused by dust from an artillery back blast event at a location in a desert region of the southwestern United States of America.
-
radial plume mapping a us epa test method for area and fugitive source emission monitoring using Optical Remote Sensing
2008Co-Authors: Ram A Hashmonay, Ravi Varma, Mark Modrak, Robert Kagann, Robin Segall, Patrick D SullivanAbstract:This paper describes the recently developed United States Environmental Protection Agency (US EPA) test method that provides the user with unique methodolo- gies for characterizing gaseous emissions from non-point pollutant sources. The radial plume mapping (RPM) methodology uses an open-path, path-integrated Optical Remote Sensing (PI-ORS) system in multiple beam configurations to directly identify emission "hot spots" and measure emission fluxes. The RPM methodology has been well devel- oped, evaluated, demonstrated, and peer reviewed. Scanning the PI-ORS system in a horizontal plane (horizontal RPM) can be used to locate hot spots of fugitive emission at ground level, while scanning in a vertical plane downwind of the area source (vertical RPM), coupled with wind measurement, can be used to measure emission fluxes. Also, scanning along a line-of-sight such as an industrial fenceline (one-dimensional RPM) can be used to profile pollutant concentrations downwind from a fugitive source. In this paper, the EPA test method is discussed, with particular reference to the RPM methodology, its applicability, limitations, and validation.
-
chapter 11 a novel method to quantify fugitive dust emissions using Optical Remote Sensing
2008Co-Authors: Ravi Varma, Ram A Hashmonay, Mark J. Rood, Ke Du, Michael R. KemmeAbstract:This paper describes a new method for retrieving path-averaged mass concentrations from multi-spectral light extinction measured by Optical Remote Sensing (ORS) instruments. The light extinction measurements as a function of wavelength were used in conjunction with an iterative inverse-Mie algorithm to retrieve path-averaged particulate matter (PM) mass distribution. Conventional mass concentration measurements in a controlled release experiment were used to calibrate the ORS method. A backscattering micro pulse lidar (MPL) was used to obtain the horizontal extent of the plume along MPL's line of sight. This method was used to measure concentrations and mass emission rates of PM with diameters ≤10 µm (PM 10 ) and PM with diameters ≤2.5 µm (PM 2.5 ) that were caused by dust from an artillery back blast event at a location in a desert region of the southwestern United States of America.
Arjan Kuijper - One of the best experts on this subject based on the ideXlab platform.
-
a light and faster regional convolutional neural network for object detection in Optical Remote Sensing images
Isprs Journal of Photogrammetry and Remote Sensing, 2018Co-Authors: Peng Ding, Ye Zhang, Weijian Deng, Ping Jia, Arjan KuijperAbstract:Abstract Detection of objects from satellite Optical Remote Sensing images is very important for many commercial and governmental applications. With the development of deep convolutional neural networks (deep CNNs), the field of object detection has seen tremendous advances. Currently, objects in satellite Remote Sensing images can be detected using deep CNNs. In general, Optical Remote Sensing images contain many dense and small objects, and the use of the original Faster Regional CNN framework does not yield a suitably high precision. Therefore, after careful analysis we adopt dense convoluted networks, a multi-scale representation and various combinations of improvement schemes to enhance the structure of the base VGG16-Net for improving the precision. We propose an approach to reduce the test-time (detection time) and memory requirements. To validate the effectiveness of our approach, we perform experiments using satellite Remote Sensing image datasets of aircraft and automobiles. The results show that the improved network structure can detect objects in satellite Optical Remote Sensing images more accurately and efficiently.