Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100% precision with recall rates of up to 60%.

SeqSLAM : visual route-based navigation for sunny summer days and stormy winter nights

Fusing satellite observations and station measurements to estimate ground-level PM2.5 is promising for monitoring PM2.5 pollution. A geo-intelligent approach, which incorporates geographical correlation into an intelligent deep learning architecture, is developed to estimate PM2.5. Specifically, it considers geographical distance and spatiotemporally correlated PM2.5 in a deep belief network (denoted as Geoi-DBN). Geoi-DBN can capture the essential features associated with PM2.5 from latent factors. It was trained and tested with data from China in 2015. The results show that Geoi-DBN performs significantly better than the traditional neural network. The out-of-sample cross-validation R2 increases from 0.42 to 0.88, and RMSE decreases from 29.96 to 13.03 μg/m3. On the basis of the derived PM2.5 distribution, it is predicted that over 80% of the Chinese population live in areas with an annual mean PM2.5 of greater than 35 μg/m3. This study provides a new perspective for air pollution monitoring in large geographic regions.

https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1002/2017GL075710

Estimating Ground-Level PM2.5 by Fusing Satellite and Station Observations: A Geo-Intelligent Deep Learning Approach

: The recent availability of geo-tagged images and rich geospatial data has inspired a number of algorithms for image based geolocalization. Most approaches predict the location of a query image by matching to ground-level images with known locations (e.g., street-view data). However, most of the Earth does not have ground-level reference photos available. Fortunately, more complete coverage is provided by oblique aerial or bird's eye imagery. In this work, we localize a ground-level query image by matching it to a reference database of aerial imagery. We use publicly available data to build a dataset of 78K aligned crossview image pairs. The primary challenge for this task is that traditional computer vision approaches cannot handle the wide baseline and appearance variation of these cross-view pairs. We use our dataset to learn a feature representation in which matching views are near one another and mismatched views are far apart. Our proposed approach, Where-CNN, is inspired by deep learning success in face verification and achieves significant improvements over traditional hand-crafted features and existing deep features learned from other large-scale databases. We show the effectiveness of Where-CNN in finding matches between street view and aerial view imagery and demonstrate the ability of our learned features to generalize to novel locations.

Learning Deep Representations for Ground to Aerial Geolocalization (Open Access)

Fusing satellite observations and station measurements to estimate ground-level PM2.5 is promising for monitoring PM2.5 pollution. A geo-intelligent approach, which incorporates geographical correlation into an intelligent deep learning architecture, is developed to estimate PM2.5. Specifically, it considers geographical distance and spatiotemporally correlated PM2.5 in a deep belief network (denoted as Geoi-DBN). Geoi-DBN can capture the essential features associated with PM2.5 from latent factors. It was trained and tested with data from China in 2015. The results show that Geoi-DBN performs significantly better than the traditional neural network. The cross-validation R increases from 0.63 to 0.94, and RMSE decreases from 29.56 to 13.68${\mu}$g/m3. On the basis of the derived PM2.5 distribution, it is predicted that over 80% of the Chinese population live in areas with an annual mean PM2.5 of greater than 35${\mu}$g/m3. This study provides a new perspective for air pollution monitoring in large geographic regions.

Estimating ground-level PM2.5 by fusing satellite and station observations: A geo-intelligent deep learning approach

An accurate and reliable Indoor Positioning System (IPS) applicable to most indoor scenarios has been sought for many years. The number of technologies, techniques, and approaches in general used in IPS proposals is remarkable. Such diversity, coupled with the lack of strict and verifiable evaluations, leads to difficulties for appreciating the true value of most proposals. This paper provides a meta-review that performed a comprehensive compilation of 62 survey papers in the area of indoor positioning. The paper provides the reader with an introduction to IPS and the different technologies, techniques, and some methods commonly employed. The introduction is supported by consensus found in the selected surveys and referenced using them. Thus, the meta-review allows the reader to inspect the IPS current state at a glance and serve as a guide for the reader to easily find further details on each technology used in IPS. The analyses of the meta-review contributed with insights on the abundance and academic significance of published IPS proposals using the criterion of the number of citations. Moreover, 75 works are identified as relevant works in the research topic from a selection of about 4000 works cited in the analyzed surveys.

https://www.mdpi.com/1424-8220/19/20/4507/pdf

A Meta-Review of Indoor Positioning Systems.

The absence of a high-quality seamless global digital elevation model (DEM) dataset has been a challenge for the Earth-related research fields. Recently, the 1-arc-second Shuttle Radar Topography Mission (SRTM-1) data have been released globally, covering over 80% of the Earth’s land surface (60°N–56°S). However, voids and anomalies still exist in some tiles, which has prevented the SRTM-1 dataset from being directly used without further processing. In this paper, we propose a method to generate a seamless DEM dataset blending SRTM-1, ASTER GDEM v2, and ICESat laser altimetry data. The ASTER GDEM v2 data are used as the elevation source for the SRTM void filling. To get a reliable filling source, ICESat GLAS points are incorporated to enhance the accuracy of the ASTER data within the void regions, using an artificial neural network (ANN) model. After correction, the voids in the SRTM-1 data are filled with the corrected ASTER GDEM values. The triangular irregular network based delta surface fill (DSF) method is then employed to eliminate the vertical bias between them. Finally, an adaptive outlier filter is applied to all the data tiles. The final result is a seamless global DEM dataset. ICESat points collected from 2003 to 2009 were used to validate the effectiveness of the proposed method, and to assess the vertical accuracy of the global DEM products in China. Furthermore, channel networks in the Yangtze River Basin were also extracted for the data assessment.

High-quality seamless DEM generation blending SRTM-1, ASTER GDEM v2 and ICESat/GLAS observations

Parsing very high resolution (VHR) urban scene images into regions with semantic meaning, e.g. buildings and cars, is a fundamental task in urban scene understanding. However, due to the huge quantity of details contained in an image and the large variations of objects in scale and appearance, the existing semantic segmentation methods often break one object into pieces, or confuse adjacent objects and thus fail to depict these objects consistently. To address these issues uniformly, we propose a standalone end-to-end edge-aware neural network (EaNet) for urban scene semantic segmentation. For semantic consistency preservation inside objects, the EaNet model incorporates a large kernel pyramid pooling (LKPP) module to capture rich multi-scale context with strong continuous feature relations. To effectively separate confusing objects with sharp contours, a Dice-based edge-aware loss function (EA loss) is devised to guide the EaNet to refine both the pixel- and image-level edge information directly from semantic segmentation prediction. In the proposed EaNet model, the LKPP and the EA loss couple to enable comprehensive feature learning across an entire semantic object. Extensive experiments on three challenging datasets demonstrate that our method can be readily generalized to multi-scale ground/aerial urban scene images, achieving 81.7% in mIoU on Cityscapes Test set and 90.8% in the mean F1-score on the ISPRS Vaihingen 2D Test set. Code is available at: https://github.com/geovsion/EaNet .

Parsing very high resolution urban scene images by learning deep ConvNets with edge-aware loss

As a result of the rapid development of smartphone-based indoor localization technology, location-based services in indoor spaces have become a topic of interest. However, to date, the rich data resulting from indoor localization and navigation applications have not been fully exploited, which is significant for trajectory correction and advanced indoor map information extraction. In this paper, an integrated location acquisition method utilizing activity recognition and semantic information extraction is proposed for indoor mobile localization. The location acquisition method combines pedestrian dead reckoning (PDR), human activity recognition (HAR) and landmarks to acquire accurate indoor localization information. Considering the problem of initial position determination, a hidden Markov model (HMM) is utilized to infer the user's initial position. To provide an improved service for further applications, the landmarks are further assigned semantic descriptions by detecting the user's activities. The experiments conducted in this study confirm that a high degree of accuracy for a user's indoor location can be obtained. Furthermore, the semantic information of a user's trajectories can be extracted, which is extremely useful for further research into indoor location applications.

Activity Recognition and Semantic Description for Indoor Mobile Localization

This paper presents a context-aware tracing strategy (CATS) for crisp edge detection with deep edge detectors, based on an observation that the localization ambiguity of deep edge detectors is mainly caused by the mixing phenomenon of convolutional neural networks: feature mixing in edge classification and side mixing during fusing side predictions. The CATS consists of two modules: a novel tracing loss that performs feature unmixing by tracing boundaries for better side edge learning, and a context-aware fusion block that tackles the side mixing by aggregating the complementary merits of learned side edges. Experiments demonstrate that the proposed CATS can be integrated into modern deep edge detectors to improve localization accuracy. With the vanilla VGG-16 backbone, in terms of BSDS dataset, our CATS improves the F-measure (ODS) of the RCF and BDCN deep edge detectors by 12% and 6% respectively when evaluating without using the morphological non-maximal suppression scheme for edge detection.

Unmixing Convolutional Features for Crisp Edge Detection.

Accurately determining pedestrian location in indoor environments using consumer smartphones is a significant step in the development of ubiquitous localization services. Many different map-matching methods have been combined with pedestrian dead reckoning (PDR) to achieve low-cost and bias-free pedestrian tracking. However, this works only in areas with dense map constraints and the error accumulates in open areas. In order to achieve reliable localization without map constraints, an improved image-based localization aided pedestrian trajectory estimation method is proposed in this paper. The image-based localization recovers the pose of the camera from the 2D-3D correspondences between the 2D image positions and the 3D points of the scene model, previously reconstructed by a structure-from-motion (SfM) pipeline. This enables us to determine the initial location and eliminate the accumulative error of PDR when an image is successfully registered. However, the image is not always registered since the traditional 2D-to-3D matching rejects more and more correct matches when the scene becomes large. We thus adopt a robust image registration strategy that recovers initially unregistered images by integrating 3D-to-2D search. In the process, the visibility and co-visibility information is adopted to improve the efficiency when searching for the correspondences from both sides. The performance of the proposed method was evaluated through several experiments and the results demonstrate that it can offer highly acceptable pedestrian localization results in long-term tracking, with an error of only 0.56 m, without the need for dedicated infrastructures.

Xianwei Zheng

Papers

High-quality seamless DEM generation blending SRTM-1, ASTER GDEM v2 and ICESat/GLAS observations

Parsing very high resolution urban scene images by learning deep ConvNets with edge-aware loss

Activity Recognition and Semantic Description for Indoor Mobile Localization

Unmixing Convolutional Features for Crisp Edge Detection.

Image-Based Localization Aided Indoor Pedestrian Trajectory Estimation Using Smartphones.