scholarly journals HD Maps: Fine-Grained Road Segmentation by Parsing Ground and Aerial Images

2016 ◽  
Author(s):  
Gellert Mattyus ◽  
Shenlong Wang ◽  
Sanja Fidler ◽  
Raquel Urtasun
Keyword(s):  
Aerial Images ◽  
Fine Grained ◽  
Road Segmentation

scholarly journals Review of Road Detection and Segmentation Technique Using CNN

2020 ◽  
pp. 161-165
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Aerial Images ◽  
Road Detection ◽  
Insurance Plan ◽  
Toll Road ◽  
White Pixel ◽  
Road Segmentation ◽  
Matlab Programming ◽  
The Relationship

This research proposes form shape mounted on “the deep convolutional neural network (CNN) for the detection of roads and the segmentation of aerial pix. Those images are received by using a UAV. The photograph segmentation set of rules has two levels: the studying segment and the working phase. The aerial images of the data deteriorated into their coloration additives, had been pre-processed in matlab on hue, after which divided into small 33 × 33 pixel packing containers the usage of a sliding container set of rules. CNN was once designed with matconvnet and had the accompanying structure: 4 convolutional levels, 4 grouping stages, a relu layer, a totally linked layer, and a softmax layer. The entire community has been organized for the use of 2,000 boxes. CNN was implemented the use of matlab programming on the gpu and the outcomes are promising. The CNN output offers pixel-by means of-pixel records, which class it has a location with (road / non-road). White pixel and choppy terrain are known as "0" (dark). Monitoring roads is a troublesome venture in aerial picture segmentation due to quite more than a few sizes and surfaces. One of the vastest steps in CNN training is the pre-processing phase. Due to toll road segmentation, dismissal structures and complexity enhancement have been applied.” this is an audited article on the relationship between representative upkeep techniques with work pleasure and responsibility in insurance plan businesses.

scholarly journals VARIABLE SELECTION FOR ROAD SEGMENTATION IN AERIAL IMAGES

2017 ◽  
Vol XLII-1/W1 ◽  
pp. 297-304 ◽  
Author(s):  
S. Warnke ◽  
D. Bulatov
Keyword(s):  
Variable Selection ◽  
Minimum Cost ◽  
Optimization Methods ◽  
Training Data ◽  
Aerial Images ◽  
Misclassification Error ◽  
3D Data ◽  
Elevation Data ◽  
Cost Paths ◽  
Road Segmentation

For extraction of road pixels from combined image and elevation data, Wegner et al. (2015) proposed classification of superpixels into road and non-road, after which a refinement of the classification results using minimum cost paths and non-local optimization methods took place. We believed that the variable set used for classification was to a certain extent suboptimal, because many variables were redundant while several features known as useful in Photogrammetry and Remote Sensing are missed. This motivated us to implement a variable selection approach which builds a model for classification using portions of training data and subsets of features, evaluates this model, updates the feature set, and terminates when a stopping criterion is satisfied. The choice of classifier is flexible; however, we tested the approach with Logistic Regression and Random Forests, and taylored the evaluation module to the chosen classifier. To guarantee a fair comparison, we kept the segment-based approach and most of the variables from the related work, but we extended them by additional, mostly higher-level features. Applying these superior features, removing the redundant ones, as well as using more accurately acquired 3D data allowed to keep stable or even to reduce the misclassification error in a challenging dataset.

scholarly journals EXPLORING SEMANTIC RELATIONSHIPS FOR HIERARCHICAL LAND USE CLASSIFICATION BASED ON CONVOLUTIONAL NEURAL NETWORKS

2020 ◽  
Vol V-2-2020 ◽  
pp. 599-607
Author(s):  
C. Yang ◽  
F. Rottensteiner ◽  
C. Heipke
Keyword(s):  
Land Use ◽  
Information Source ◽  
Semantic Category ◽  
Semantic Segmentation ◽  
Aerial Images ◽  
Land Use Classification ◽  
Semantic Relationships ◽  
Fine Grained ◽  
Semantic Dependencies ◽  
Number Of Classes

Abstract. Land use (LU) is an important information source commonly stored in geospatial databases. Most current work on automatic LU classification for updating topographic databases considers only one category level (e.g. residential or agricultural) consisting of a small number of classes. However, LU databases frequently contain very detailed information, using a hierarchical object catalogue where the number of categories differs depending on the hierarchy level. This paper presents a method for the classification of LU on the basis of aerial images that differentiates a fine-grained class structure, exploiting the hierarchical relationship between categories at different levels of the class catalogue. Starting from a convolutional neural network (CNN) for classifying the categories of all levels, we propose a strategy to simultaneously learn the semantic dependencies between different category levels explicitly. The input to the CNN consists of aerial images and derived data as well as land cover information derived from semantic segmentation. Its output is the class scores at three different semantic levels, based on which predictions that are consistent with the class hierarchy are made. We evaluate our method using two test sites and show how the classification accuracy depends on the semantic category level. While at the coarsest level, an overall accuracy in the order of 90% can be achieved, at the finest level, this accuracy is reduced to around 65%. Our experiments also show which classes are particularly hard to differentiate.

scholarly journals Automated Method of Road Extraction from Aerial Images Using a Deep Convolutional Neural Network

2019 ◽  
Vol 9 (22) ◽  
pp. 4825 ◽  
Author(s):  
Tamara Alshaikhli ◽  
Wen Liu ◽  
Yoshihisa Maruyama
Keyword(s):  
Traffic Management ◽  
City Planning ◽  
Road Networks ◽  
Good Alternative ◽  
Aerial Images ◽  
Training Dataset ◽  
Automated Method ◽  
Learning Techniques ◽  
Manual Methods ◽  
Road Segmentation

Updating road networks using remote sensing imagery is among the most important topics in city planning, traffic management and disaster management. As a good alternative to manual methods, which are considered to be expensive and time consuming, deep learning techniques provide great improvements in these regards. One of these techniques is the use of deep convolution neural networks (DCNNs). This study presents a road segmentation model consisting of a skip connection of U-net and residual blocks (ResBlocks) in the encoding part and convolution layers (Conv. layer) in the decoding part. Although the model uses fewer residual blocks in the encoding part and fewer convolution layers in the decoding part, it produces better image predictions in comparison with other state-of-the-art models. This model automatically and efficiently extracts road networks from high-resolution aerial imagery in an unexpansive manner using a small training dataset.

scholarly journals Multi-resolution dataset for photovoltaic panel segmentation from satellite and aerial imagery

2021 ◽  
Vol 13 (11) ◽  
pp. 5389-5401
Author(s):  
Hou Jiang ◽  
Ling Yao ◽  
Ning Lu ◽  
Jun Qin ◽  
Tang Liu ◽  
...  
Keyword(s):  
Rapid Development ◽  
Model Performance ◽  
Detection Algorithm ◽  
Fine Tuning ◽  
Aerial Images ◽  
Photovoltaic Panel ◽  
Fine Grained ◽  
Automatic Information ◽  
Deep Networks ◽  
Direct Cross

Abstract. In the context of global carbon emission reduction, solar photovoltaic (PV) technology is experiencing rapid development. Accurate localized PV information, including location and size, is the basis for PV regulation and potential assessment of the energy sector. Automatic information extraction based on deep learning requires high-quality labeled samples that should be collected at multiple spatial resolutions and under different backgrounds due to the diversity and variable scale of PVs. We established a PV dataset using satellite and aerial images with spatial resolutions of 0.8, 0.3, and 0.1 m, which focus on concentrated PVs, distributed ground PVs, and fine-grained rooftop PVs, respectively. The dataset contains 3716 samples of PVs installed on shrub land, grassland, cropland, saline–alkali land, and water surfaces, as well as flat concrete, steel tile, and brick roofs. The dataset is used to examine the model performance of different deep networks on PV segmentation. On average, an intersection over union (IoU) greater than 85 % is achieved. In addition, our experiments show that direct cross application between samples with different resolutions is not feasible and that fine-tuning of the pre-trained deep networks using target samples is necessary. The dataset can support more work on PV technology for greater value, such as developing a PV detection algorithm, simulating PV conversion efficiency, and estimating regional PV potential. The dataset is available from Zenodo on the following website: https://doi.org/10.5281/zenodo.5171712 (Jiang et al., 2021).

scholarly journals Multi-resolution dataset for photovoltaic panel segmentation from satellite and aerial imagery

2021 ◽  
Author(s):  
Hou Jiang ◽  
Ling Yao ◽  
Ning Lu ◽  
Jun Qin ◽  
Tang Liu ◽  
...  
Keyword(s):  
Rapid Development ◽  
Model Performance ◽  
Detection Algorithm ◽  
Fine Tuning ◽  
Aerial Images ◽  
Photovoltaic Panel ◽  
Fine Grained ◽  
Automatic Information ◽  
Deep Networks ◽  
Direct Cross

Abstract. In the context of global carbon emission reduction, solar photovoltaics (PV) is experiencing rapid development. Accurate localized PV information, including location and size, is the basis for PV regulation and potential assessment of energy sector. Automatic information extraction based on deep learning requires high-quality labelled samples that should be collected at multiple spatial resolutions and under different backgrounds due to the diversity and variable scale of PV. We established a PV dataset using satellite and aerial images with spatial resolutions of 0.8 m, 0.3 m and 0.1 m, which focus on concentrated PV, distributed ground PV and fine-grained rooftop PV, respectively. The dataset contains 3716 samples of PVs installed on shrub land, grassland, cropland, saline-alkali, and water surface, as well as flat concrete, steel tile, and brick roofs. We used this dataset to examine the model performance of different deep networks on PV segmentation, and on average an intersection over union (IoU) greater than 85 % was achieved. In addition, our experiments show that direct cross application between samples with different resolutions is not feasible, and fine-tuning of the pre-trained deep networks using target samples is necessary. The dataset can support more works on PVs for greater value, such as, developing PV detection algorithm, simulating PV conversion efficiency, and estimating regional PV potential. The dataset is available from Zenodo on the following website: https://doi.org/10.5281/zenodo.5171712 (Jiang et al. 2021).

Controls of channel morphodynamics on the intertidal dune morphodynamics and associated bedload transport in the open-coast macrotidal flats

2020 ◽  
Author(s):  
Joohee Jo ◽  
Dohyeong Kim ◽  
Kyungsik Choi
Keyword(s):  
Bedload Transport ◽  
Aerial Images ◽  
Point Bar ◽  
Fine Grained ◽  
Open Coast ◽  
Lower Intertidal Zone ◽  
Downslope Flow ◽  
Runoff Discharge ◽  
Research Period

<p>Intertidal dune morphodynamics is closely tied to bedload transport that is variable in time and space due to the interplay between tide, wave and runoff discharge. Surprisingly the control of intertidal channel morphology on the dune morphodynamics and related bedload transport is scarcely documented. Actively migrating dunes are widely developed in the lower intertidal zone of Yeochari tidal flat in the northern Gyeonggi Bay, west coast of Korea. High-resolution aerial images, high-precision transect profiles, and hydrodynamic dataset were repeatedly obtained and analyzed to quantify the intertidal dune morphodynamics and associated bedload transport, and to address the role of channel morphodynamics. During the research period, the intertidal channel became more sinuous and an ebb barb arose concurrently at the upstream of the channel point bar. The ebb barb exerted a key role in the downstream delivery of fine-grained sediments onto the areas covered by dunes and the intertidal channel by reinforcing ebb currents with a pronounced time-velocity asymmetry. The presence of the ebb barb resulted in a rapid decrease of the width/depth ratio of the channel that had migrated laterally 130 m in six years. After the ebb-barb development, the heights and steepness (height/wavelength) of dunes on the point bar and near the ebb barb decreased notably. Simultaneously dune migration rate had increased from 0.5 m/day to 2.5 m/day, which decreases away from the channel. Bedload transport estimated by using Meyer-Peter and Muller (MPM) equation and Dune-Tracking Method (DTM) also decreases away from the channel. Bedload transport calculated by DTM (qb<sub>DTM</sub>, 0.03-0.38 m<sup>2</sup>/day) is much smaller than that estimated by MPM (qb<sub>MPM</sub>, 0.10-4.17 m<sup>2</sup>/day) by a factor of 1.5 to 62. The discrepancy ratio between the two bedload estimates (qb<sub>MPM</sub>/qb<sub>DTM</sub>) increases toward the channel and the ebb barb. Downslope flow toward the channel during the late stage of ebb tide may account for the underestimation of qb<sub>DTM</sub> by facilitating downslope sediment transport that reduced the dune steepness with the infilling of dune trough. The present study showcased a dynamic response of the dune morphodynamics and associated bedload transport in the open-coast tidal flats to the changes in the channel morphodynamics that is controlled by seasonal runoff discharge as well as tidal currents.</p>

Sign in / Sign up

Export Citation Format

Share Document