scholarly journals BUILDING CHANGE DETECTION FROM BITEMPORAL AERIAL IMAGES USING DEEP LEARNING

2020 ◽  
Vol V-2-2020 ◽  
pp. 565-571
Author(s):  
S. Su ◽  
T. Nawata ◽  
T. Fuse
Keyword(s):  
Deep Learning ◽  
Change Detection ◽  
Urban Areas ◽  
Network Architecture ◽  
Aerial Images ◽  
Detection Accuracy ◽  
Detection Model ◽  
Wide Range ◽  
Surface Models ◽  
Data Source

Abstract. Automatic building change detection has become a topical issue owing to its wide range of applications, such as updating building maps. However, accurate building change detection remains challenging, particularly in urban areas. Thus far, there has been limited research on the use of the outdated building map (the building map before the update, referred to herein as the old-map) to increase the accuracy of building change detection. This paper presents a novel deep-learning-based method for building change detection using bitemporal aerial images containing RGB bands, bitemporal digital surface models (DSMs), and an old-map. The aerial images have two types of spatial resolutions, 12.5 cm or 16 cm, and the cell size of the DSMs is 50 cm × 50 cm. The bitemporal aerial images, the height variations calculated using the differences between the bitemporal DSMs, and the old-map were fed into a network architecture to build an automatic building change detection model. The performance of the model was quantitatively and qualitatively evaluated for an urban area that covered approximately 10 km2 and contained over 21,000 buildings. The results indicate that it can detect the building changes with optimum accuracy as compared to other methods that use inputs such as i) bitemporal aerial images only, ii) bitemporal aerial images and bitemporal DSMs, and iii) bitemporal aerial images and an old-map. The proposed method achieved recall rates of 89.3%, 88.8%, and 99.5% for new, demolished, and other buildings, respectively. The results also demonstrate that the old-map is an effective data source for increasing building change detection accuracy.

scholarly journals Building Instance Change Detection from Large-Scale Aerial Images using Convolutional Neural Networks and Simulated Samples

2019 ◽  
Vol 11 (11) ◽  
pp. 1343 ◽  
Author(s):  
Shunping Ji ◽  
Yanyun Shen ◽  
Meng Lu ◽  
Yongjun Zhang
Keyword(s):  
Deep Learning ◽  
Change Detection ◽  
Urban Areas ◽  
Large Scale ◽  
Real Life ◽  
Aerial Images ◽  
Building Extraction ◽  
Object Based ◽  
Wide Range ◽  
Change Map

We present a novel convolutional neural network (CNN)-based change detection framework for locating changed building instances as well as changed building pixels from very high resolution (VHR) aerial images. The distinctive advantage of the framework is the self-training ability, which is highly important in deep-learning-based change detection in practice, as high-quality samples of changes are always lacking for training a successful deep learning model. The framework consists two parts: a building extraction network to produce a binary building map and a building change detection network to produce a building change map. The building extraction network is implemented with two widely used structures: a Mask R-CNN for object-based instance segmentation, and a multi-scale full convolutional network for pixel-based semantic segmentation. The building change detection network takes bi-temporal building maps produced from the building extraction network as input and outputs a building change map at the object and pixel levels. By simulating arbitrary building changes and various building parallaxes in the binary building map, the building change detection network is well trained without real-life samples. This greatly lowers the requirements of labeled changed buildings, and guarantees the algorithm’s robustness to registration errors caused by parallaxes. To evaluate the proposed method, we chose a wide range of urban areas from an open-source dataset as training and testing areas, and both pixel-based and object-based model evaluation measures were used. Experiments demonstrated our approach was vastly superior: without using any real change samples, it reached 63% average precision (AP) at the object (building instance) level. In contrast, with adequate training samples, other methods—including the most recent CNN-based and generative adversarial network (GAN)-based ones—have only reached 25% AP in their best cases.

scholarly journals Deep Learning-Based Thermal Image Analysis for Pavement Defect Detection and Classification Considering Complex Pavement Conditions

2021 ◽  
Vol 14 (1) ◽  
pp. 106
Author(s):  
Cheng Chen ◽  
Sindhu Chandra ◽  
Yufan Han ◽  
Hyungjoon Seo
Keyword(s):  
Deep Learning ◽  
Damage Detection ◽  
Defect Detection ◽  
Input Data ◽  
Thermal Image ◽  
Detection Accuracy ◽  
Detection Model ◽  
Fused Image ◽  
Data Source ◽  
Pavement Damage

Automatic damage detection using deep learning warrants an extensive data source that captures complex pavement conditions. This paper proposes a thermal-RGB fusion image-based pavement damage detection model, wherein the fused RGB-thermal image is formed through multi-source sensor information to achieve fast and accurate defect detection including complex pavement conditions. The proposed method uses pre-trained EfficientNet B4 as the backbone architecture and generates an argument dataset (containing non-uniform illumination, camera noise, and scales of thermal images too) to achieve high pavement damage detection accuracy. This paper tests separately the performance of different input data (RGB, thermal, MSX, and fused image) to test the influence of input data and network on the detection results. The results proved that the fused image’s damage detection accuracy can be as high as 98.34% and by using the dataset after augmentation, the detection model deems to be more stable to achieve 98.35% precision, 98.34% recall, and 98.34% F1-score.

scholarly journals Aerial and UAV Images for Photogrammetric Analysis of Belvedere Glacier Evolution in the Period 1977–2019

2021 ◽  
Vol 13 (18) ◽  
pp. 3787
Author(s):  
Carlo Iapige De De Gaetani ◽  
Francesco Ioli ◽  
Livio Pinto
Keyword(s):  
Aerial Images ◽  
Full Potential ◽  
Alpine Glacier ◽  
The Past ◽  
Long Span ◽  
Wide Range ◽  
Surface Models ◽  
Aerial Platforms ◽  
Uav Images ◽  
Modern Technologies

Alpine glaciers are strongly suffering the consequences of the temperature rising and monitoring them over long periods is of particular interest for climate change tracking. A wide range of techniques can be successfully applied to survey and monitor glaciers with different spatial and temporal resolutions. However, going back in time to retrace the evolution of a glacier is still a challenging task. Historical aerial images, e.g., those acquired for regional cartographic purposes, are extremely valuable resources for studying the evolution and movement of a glacier in the past. This work analyzed the evolution of the Belvedere Glacier by means of structure from motion techniques applied to digitalized historical aerial images combined with more recent digital surveys, either from aerial platforms or UAVs. This allowed the monitoring of an Alpine glacier with high resolution and geometrical accuracy over a long span of time, covering the period 1977-2019. In this context, digital surface models of the area at different epochs were computed and jointly analyzed, retrieving the morphological dynamics of the Belvedere Glacier. The integration of datasets dating back to earlier times with those referring to surveys carried out with more modern technologies exploits at its full potential the information that at first glance could be thought obsolete, proving how historical photogrammetric datasets are a remarkable heritage for glaciological studies.

scholarly journals A Deep Learning-Based Approach for the Detection of Early Signs of Gingivitis in Orthodontic Patients Using Faster Region-Based Convolutional Neural Networks

2020 ◽  
Vol 17 (22) ◽  
pp. 8447
Author(s):  
Dima M. Alalharith ◽  
Hajar M. Alharthi ◽  
Wejdan M. Alghamdi ◽  
Yasmine M. Alsenbel ◽  
Nida Aslam ◽  
...  
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Periodontal Disease ◽  
Convolutional Neural Network ◽  
Region Of Interest ◽  
Training Dataset ◽  
Detection Accuracy ◽  
Detection Model ◽  
Art Object ◽  
Orthodontic Patients

Computer-based technologies play a central role in the dentistry field, as they present many methods for diagnosing and detecting various diseases, such as periodontitis. The current study aimed to develop and evaluate the state-of-the-art object detection and recognition techniques and deep learning algorithms for the automatic detection of periodontal disease in orthodontic patients using intraoral images. In this study, a total of 134 intraoral images were divided into a training dataset (n = 107 [80%]) and a test dataset (n = 27 [20%]). Two Faster Region-based Convolutional Neural Network (R-CNN) models using ResNet-50 Convolutional Neural Network (CNN) were developed. The first model detects the teeth to locate the region of interest (ROI), while the second model detects gingival inflammation. The detection accuracy, precision, recall, and mean average precision (mAP) were calculated to verify the significance of the proposed model. The teeth detection model achieved an accuracy, precision, recall, and mAP of 100 %, 100%, 51.85%, and 100%, respectively. The inflammation detection model achieved an accuracy, precision, recall, and mAP of 77.12%, 88.02%, 41.75%, and 68.19%, respectively. This study proved the viability of deep learning models for the detection and diagnosis of gingivitis in intraoral images. Hence, this highlights its potential usability in the field of dentistry and aiding in reducing the severity of periodontal disease globally through preemptive non-invasive diagnosis.

scholarly journals Evaluation of Unsupervised Change Detection Methods Applied to Landslide Inventory Mapping Using ASTER Imagery

2018 ◽  
Vol 10 (12) ◽  
pp. 1987 ◽  
Author(s):  
Rocío Ramos-Bernal ◽  
René Vázquez-Jiménez ◽  
Raúl Romero-Calcerrada ◽  
Patricia Arrogante-Funes ◽  
Carlos Novillo
Keyword(s):  
Change Detection ◽  
Vegetation Index ◽  
Thermal Emission ◽  
Principal Component ◽  
Detection Methods ◽  
Detection Accuracy ◽  
Data Types ◽  
Statistical Parameters ◽  
Wide Range ◽  
The World

Natural hazards include a wide range of high-impact phenomena that affect socioeconomic and natural systems. Landslides are a natural hazard whose destructive power has caused a significant number of victims and substantial damage around the world. Remote sensing provides many data types and techniques that can be applied to monitor their effects through landslides inventory maps. Three unsupervised change detection methods were applied to the Advanced Spaceborne Thermal Emission and Reflection Radiometer (Aster)-derived images from an area prone to landslides in the south of Mexico. Linear Regression (LR), Chi-Square Transformation, and Change Vector Analysis were applied to the principal component and the Normalized Difference Vegetation Index (NDVI) data to obtain the difference image of change. The thresholding was performed on the change histogram using two approaches: the statistical parameters and the secant method. According to previous works, a slope mask was used to classify the pixels as landslide/No-landslide; a cloud mask was used to eliminate false positives; and finally, those landslides less than 450 m2 (two Aster pixels) were discriminated. To assess the landslide detection accuracy, 617 polygons (35,017 pixels) were sampled, classified as real landslide/No-landslide, and defined as ground-truth according to the interpretation of color aerial photo slides to obtain omission/commission errors and Kappa coefficient of agreement. The results showed that the LR using NDVI data performs the best results in landslide detection. Change detection is a suitable technique that can be applied for the landslides mapping and we think that it can be replicated in other parts of the world with results similar to those obtained in the present work.

scholarly journals Detecting Objects from Space: An Evaluation of Deep-Learning Modern Approaches

Electronics ◽  
2020 ◽  
Vol 9 (4) ◽  
pp. 583 ◽  
Author(s):  
Khang Nguyen ◽  
Nhut T. Huynh ◽  
Phat C. Nguyen ◽  
Khanh-Duy Nguyen ◽  
Nguyen D. Vo ◽  
...  
Keyword(s):  
Deep Learning ◽  
Object Detection ◽  
Unmanned Aircraft ◽  
Aerial Images ◽  
Great Success ◽  
Single Shot ◽  
Convolutional Networks ◽  
Image Pyramids ◽  
Fully Convolutional Networks ◽  
Wide Range

Unmanned aircraft systems or drones enable us to record or capture many scenes from the bird’s-eye view and they have been fast deployed to a wide range of practical domains, i.e., agriculture, aerial photography, fast delivery and surveillance. Object detection task is one of the core steps in understanding videos collected from the drones. However, this task is very challenging due to the unconstrained viewpoints and low resolution of captured videos. While deep-learning modern object detectors have recently achieved great success in general benchmarks, i.e., PASCAL-VOC and MS-COCO, the robustness of these detectors on aerial images captured by drones is not well studied. In this paper, we present an evaluation of state-of-the-art deep-learning detectors including Faster R-CNN (Faster Regional CNN), RFCN (Region-based Fully Convolutional Networks), SNIPER (Scale Normalization for Image Pyramids with Efficient Resampling), Single-Shot Detector (SSD), YOLO (You Only Look Once), RetinaNet, and CenterNet for the object detection in videos captured by drones. We conduct experiments on VisDrone2019 dataset which contains 96 videos with 39,988 annotated frames and provide insights into efficient object detectors for aerial images.

scholarly journals Classification of Shellfish Recognition Based on Improved Faster R-CNN Framework of Deep Learning

2021 ◽  
Vol 2021 ◽  
pp. 1-10
Author(s):  
Yiran Feng ◽  
Xueheng Tao ◽  
Eung-Joo Lee
Keyword(s):  
Deep Learning ◽  
Learning Algorithm ◽  
Experimental Tests ◽  
Detection Algorithm ◽  
Detection Accuracy ◽  
Multiple Objects ◽  
Detection Model ◽  
Deep Learning Algorithm ◽  
Quality Sorting

In view of the current absence of any deep learning algorithm for shellfish identification in real contexts, an improved Faster R-CNN-based detection algorithm is proposed in this paper. It achieves multiobject recognition and localization through a second-order detection network and replaces the original feature extraction module with DenseNet, which can fuse multilevel feature information, increase network depth, and avoid the disappearance of network gradients. Meanwhile, the proposal merging strategy is improved with Soft-NMS, where an attenuation function is designed to replace the conventional NMS algorithm, thereby avoiding missed detection of adjacent or overlapping objects and enhancing the network detection accuracy under multiple objects. By constructing a real contexts shellfish dataset and conducting experimental tests on a vision recognition seafood sorting robot production line, we were able to detect the features of shellfish in different scenarios, and the detection accuracy was improved by nearly 4% compared to the original detection model, achieving a better detection accuracy. This provides favorable technical support for future quality sorting of seafood using the improved Faster R-CNN-based approach.

scholarly journals Co-Segmentation and Superpixel-Based Graph Cuts for Building Change Detection from Bi-Temporal Digital Surface Models and Aerial Images

2019 ◽  
Vol 11 (6) ◽  
pp. 729 ◽  
Author(s):  
Shiyan Pang ◽  
Xiangyun Hu ◽  
Mi Zhang ◽  
Zhongliang Cai ◽  
Fengzhu Liu
Keyword(s):  
Change Detection ◽  
Point Cloud ◽  
Laser Scanner ◽  
Graph Cuts ◽  
Aerial Images ◽  
Processing Unit ◽  
3D Point Cloud ◽  
Point Cloud Data ◽  
Cloud Data ◽  
Surface Models

Thanks to the recent development of laser scanner hardware and the technology of dense image matching (DIM), the acquisition of three-dimensional (3D) point cloud data has become increasingly convenient. However, how to effectively combine 3D point cloud data and images to realize accurate building change detection is still a hotspot in the field of photogrammetry and remote sensing. Therefore, with the bi-temporal aerial images and point cloud data obtained by airborne laser scanner (ALS) or DIM as the data source, a novel building change detection method combining co-segmentation and superpixel-based graph cuts is proposed in this paper. In this method, the bi-temporal point cloud data are firstly combined to achieve a co-segmentation to obtain bi-temporal superpixels with the simple linear iterative clustering (SLIC) algorithm. Secondly, for each period of aerial images, semantic segmentation based on a deep convolutional neural network is used to extract building areas, and this is the basis for subsequent superpixel feature extraction. Again, with the bi-temporal superpixel as the processing unit, a graph-cuts-based building change detection algorithm is proposed to extract the changed buildings. In this step, the building change detection problem is modeled as two binary classifications, and acquisition of each period’s changed buildings is a binary classification, in which the changed building is regarded as foreground and the other area as background. Then, the graph cuts algorithm is used to obtain the optimal solution. Next, by combining the bi-temporal changed buildings and digital surface models (DSMs), these changed buildings are further classified as “newly built,” “taller,” “demolished”, and “lower”. Finally, two typical datasets composed of bi-temporal aerial images and point cloud data obtained by ALS or DIM are used to validate the proposed method, and the experiments demonstrate the effectiveness and generality of the proposed algorithm.

scholarly journals COMPARING ICESAT/GLAS BASED ELEVATION HEIGHTS WITH PHOTOGRAMMETRIC TERRAIN HEIGHTS FROM UAV-IMAGERY ON THE EAST TIBETAN PLATEAU

2015 ◽  
Vol XL-3/W3 ◽  
pp. 385-390 ◽  
Author(s):  
F. Enßle ◽  
A. Fritz ◽  
B. Koch
Keyword(s):  
Tibetan Plateau ◽  
Digital Elevation Models ◽  
Shuttle Radar Topography Mission ◽  
Aerial Images ◽  
Laser Altimeter ◽  
Ice Cloud ◽  
Digital Elevation ◽  
Surface Models ◽  
Elevation Data ◽  
Data Source

Digital elevation models (DEMs) and height measurements are broadly used in environmental studies. Two common elevation sources are the Ice Cloud and land elevation Satellite (ICESat), which acquired laser range measurements with the Geoscience Laser Altimeter System (GLAS) across the globe and elevation data from the Shuttle Radar Topography Mission (SRTM). Current developments of small unmanned aerial vehicles (UAV) provide the opportunity to collect aerial images of remote areas at a high spatial resolution. These can be further processed to digital surface models by stereophotogrammetry and provide a reliable data source to evaluate coarse scale Digital Elevation Models (DEMs). <br><br> This study compares ICESat/GLAS and SRTM90 elevation data against photogrammetric terrain heights within GLAS footprints on high altitudes on the East Tibetan Plateau. Without vegetation-bias, we were able to examine height differences under different topographic conditions and of different acquisition dates. Several resampling techniques were applied to SRTM90 data and averaged height within each footprint was calculated. ICESat/GLAS heights (n = 148) are most similar to UAV data based elevations with an averaged difference of &minus;0.8m ±3.1m. Results furthermore indicate the validity of ICESat/GLAS heights, which are usually removed from analyses by applying different quality flags. Smallest difference of SRTM90 to UAV based heights could be observed by a natural neighbour resampling technique (averaged 3.6m ±14m), whereat other techniques achieved quite similar results. It can be confirmed that within a range of 3,800&ndash;4,200m above mean sea level the ICESat/GLAS heights are a precise source to determine elevation at footprint geolocation.

scholarly journals Stacking Ensemble Methodology Using Deep Learning and ARIMA Models for Short-Term Load Forecasting

Energies ◽  
2021 ◽  
Vol 14 (21) ◽  
pp. 7378
Author(s):  
Pedro M. R. Bento ◽  
Jose A. N. Pombo ◽  
Maria R. A. Calado ◽  
Silvio J. P. S. Mariano
Keyword(s):  
Deep Learning ◽  
New England ◽  
Network Architecture ◽  
Prediction Models ◽  
Load Forecasting ◽  
Training Data ◽  
Constant Field ◽  
Short Term ◽  
Wide Range ◽  
Short Term Load Forecasting

Short-Term Load Forecasting is critical for reliable power system operation, and the search for enhanced methodologies has been a constant field of investigation, particularly in an increasingly competitive environment where the market operator and its participants need to better inform their decisions. Hence, it is important to continue advancing in terms of forecasting accuracy and consistency. This paper presents a new deep learning-based ensemble methodology for 24 h ahead load forecasting, where an automatic framework is proposed to select the best Box-Jenkins models (ARIMA Forecasters), from a wide-range of combinations. The method is distinct in its parameters but more importantly in considering different batches of historical (training) data, thus benefiting from prediction models focused on recent and longer load trends. Afterwards, these accurate predictions, mainly the linear components of the load time-series, are fed to the ensemble Deep Forward Neural Network. This flexible type of network architecture not only functions as a combiner but also receives additional historical and auxiliary data to further its generalization capabilities. Numerical testing using New England market data validated the proposed ensemble approach with diverse base forecasters, achieving promising results in comparison with other state-of-the-art methods.

Sign in / Sign up

Export Citation Format

Share Document