Aerial imagery analysis is critical for many research fields. However, obtaining frequent high-quality aerial images is not always accessible due to its high effort and cost requirements. One solution is to use the Ground-to-Aerial (G2A) technique to synthesize aerial images from easily collectible ground images. However, G2A is rarely studied, because of its challenges, including but not limited to, the drastic view changes, occlusion, and range of visibil-ity. In this paper, we present a novel Geometric Preserving Ground-to-Aerial (G2A) image synthesis (GPG2A) model that can generate realistic aerial images from ground images. GPG2A consists of two stages. The first stage predicts the Bird's Eye View (BEV) segmentation (referred to as the BEV layout map) from the ground image. The second stage synthesizes the aerial image from the predicted BEV layout map and text descriptions of the ground image. To train our model, we present a new multimodal cross-view dataset, namely VIGORv2, built upon VIGOR [64] with newly collected aerial images, maps, and text descriptions. Our extensive experiments illustrate that GPG2A synthesizes better geometry-preserved aerial images than existing models. We also present two applications, data augmentation for cross-view geo-localization and sketch-based region search, to further verify the effectiveness of our GPG2A. The code and dataset are available at https://github.com/AhmadArrabi/GPG2A.

Sensor nodes’ localization in wireless Internet of Things (IoT) sensor networks is crucial for the effective operation of diverse applications, such as smart cities and smart agriculture. Existing sensor nodes’ localization approaches heavily rely on anchor nodes within wireless sensor networks (WSNs). Anchor nodes are sensor nodes equipped with global positioning system (GPS) receivers and thus, have known locations. These anchor nodes operate as references to localize other sensor nodes. However, the presence of anchor nodes may not always be feasible in real-world IoT scenarios. Additionally, localization accuracy can be compromised by fluctuations in received signal strength indicator (RSSI), particularly under non-line-of-sight (NLOS) conditions. To address these challenges, we propose UBiGTLoc, a Unified bidirectional long-short-term memory (BiLSTM)-Graph Transformer Localization framework. The proposed UBiGTLoc framework effectively localizes sensor nodes in both anchor-free and anchor-presence WSNs. The framework leverages BiLSTM networks to capture temporal variations in RSSI data and employs Graph Transformer layers to model spatial relationships between sensor nodes. Extensive simulations demonstrate that UBiGTLoc consistently outperforms existing methods and provides robust localization across both dense and sparse WSNs while relying solely on cost-effective RSSI data.

Cross-View Geo-Localization (CVGL) estimates the location of a ground image by matching it to a geo-tagged aerial image in a database. Recent works achieve outstanding progress on CVGL benchmarks. However, existing methods still suffer from poor performance in cross-area evaluation, in which the training and testing data are captured from completely distinct areas. We attribute this deficiency to the lack of ability to extract the geometric layout of visual features and models' overfitting to low-level details. Our preliminary work (Zhang et al. 2022) introduced a Geometric Layout Extractor (GLE) to capture the geometric layout from input features. However, the previous GLE does not fully exploit information in the input feature. In this work, we propose GeoDTR+ with an enhanced GLE module that better models the correlations among visual features. To fully explore the LS techniques from our preliminary work, we further propose Contrastive Hard Samples Generation (CHSG) to facilitate model training. Extensive experiments show that GeoDTR+ achieves state-of-the-art (SOTA) results in cross-area evaluation on CVUSA (Workman et al. 2015), CVACT (Liu and Li, 2019), and VIGOR (Zhu et al. 2021) by a large margin (16.44%, 22.71%, and 13.66% without polar transformation) while keeping the same-area performance comparable to existing SOTA. Moreover, we provide detailed analyses of GeoDTR+.

Cross-view geo-localization aims to estimate the location of a query ground image by matching it to a reference geo-tagged aerial images database. As an extremely challenging task, its difficulties root in the drastic view changes and different capturing time between two views. Despite these difficulties, recent works achieve outstanding progress on cross-view geo-localization benchmarks. However, existing methods still suffer from poor performance on the cross-area benchmarks, in which the training and testing data are captured from two different regions. We attribute this deficiency to the lack of ability to extract the spatial configuration of visual feature layouts and models' overfitting on low-level details from the training set. In this paper, we propose GeoDTR which explicitly disentangles geometric information from raw features and learns the spatial correlations among visual features from aerial and ground pairs with a novel geometric layout extractor module. This module generates a set of geometric layout descriptors, modulating the raw features and producing high-quality latent representations. In addition, we elaborate on two categories of data augmentations, (i) Layout simulation, which varies the spatial configuration while keeping the low-level details intact. (ii) Semantic augmentation, which alters the low-level details and encourages the model to capture spatial configurations. These augmentations help to improve the performance of the cross-view geo-localization models, especially on the cross-area benchmarks. Moreover, we propose a counterfactual-based learning process to benefit the geometric layout extractor in exploring spatial information. Extensive experiments show that GeoDTR not only achieves state-of-the-art results but also significantly boosts the performance on same-area and cross-area benchmarks. Our code can be found at https://gitlab.com/vail-uvm/geodtr.

Cross-view geo-localization aims to estimate the GPS location of a query ground-view image by matching it to images from a reference database of geo-tagged aerial images. To address this challenging problem, recent approaches use panoramic ground-view images to increase the range of visibility. Although appealing, panoramic images are not readily available compared to the videos of limited Field-Of-View (FOV) images. In this paper, we present the first cross-view geo-localization method that works on a sequence of limited FOV images. Our model is trained end-to-end to capture the temporal structure that lies within the frames using the attention-based temporal feature aggregation module. To robustly tackle different sequences length and GPS noises during inference, we propose to use a sequential dropout scheme to simulate variant length sequences. To evaluate the proposed approach in realistic settings, we present a new large-scale dataset containing ground-view sequences along with the corresponding aerial-view images. Extensive experiments and comparisons demonstrate the superiority of the proposed approach compared to several competitive baselines.

For safety purposes, understanding the quality of the lane marks is essential for advanced driving technologies and pavement road maintenance. In practice, the performance of existing lane detection algorithms [1]–[3] struggles with lane marks in poor conditions, especially in rural areas where the lane marks are less maintained. However, enhancing the quality of lane marks is still not thoroughly studied. In this research, we collected a new public dataset to benchmark the performance of lane detection algorithms on various lane marks conditions from different rural areas in the U.S. Also, this paper proposed LanePainter, a Generative Adversarial Network (GAN) based model, which simultaneously enhances and assesses lane marks quality in street view images. To increase the detectability of lane marks, LanePainter enhances degraded lane marks of street view images directly on image pixels without affecting the high-quality lane marks. Our experiments demonstrated a substantial increase in the performance of current lane detection algorithms on low-quality lane marks after enhancing the images using our proposed LanePainter algorithm. Also, pavement maintenance can take benefit from our classification assessment method to quickly and accurately locate the regions with low-quality lane marks.

Traffic signs recognition (TSR) is a crucial sub-domain of computer vision, particularly relevant to autonomous vehicles and autonomous driver-assistance systems (ADAS). TSR systems can further augment efforts in many other applications, such as highway asset maintenance and management. Despite the relative success of detectors with hand-crafted features in Europe, most, if not all, non-deep-learning based systems are not scalable to accurately recognize a large subset of traffic signs in the United States. The recent works in the domain of object detection using machine learning have shown the necessity of deep neural networks (DNNs) in TSR, whereby a DNN can learn features automatically without hand-crafting them. Due to the lack of datasets in the U.S. and the inefficient use of traditional methods for traffic sign recognition (TSR) in the U.S., we created the Automotive Repository of Traffic Signs (ARTS), a new dataset for traffic signs recognition in the U.S. ARTS covers a wide range of sign-types, including Regulatory, Guide, Warning, and Temporary signs as defined in the Manual on Uniform Traffic Control Devices (MUTCD). It also features geospatial data to localize signs using their GPS coordinates. Benchmarks are presented to assess the performance of state-of-the-art deep learning based detectors.

Power system models need regular calibration to reflect the current system status, identify and mitigate potential issues. However, traditional calibration approaches usually have local minima issues that require intervention from domain experts. In this study, we propose a reinforcement learning algorithm based on hierarchical parameter tuning (HPT-RL) for power system model calibration. The proposed HPT-RL is inspired by the natural hierarchy in parameter tuning where its higher level learns which parameter to tune and its lower level learns the tuning step size. We implement HPT-RL using the Deep Deterministic Policy Gradient (DDPG) architecture. We evaluate and compare the performance of HPT-RL, traditional DDPG, Bayesian Optimization, and Random Search for a generator model calibration on a four-bus system simulated with PSS®E. The performance of HPT-RL was evaluated based on 100 repeated experiments for two scenarios, namely low randomization and high randomization, based on where a simulated disturbance occurs on the bus system. The results show that HPT-RL outperforms the other three benchmark methods.

This paper focuses on detecting the respiratory of a person by utilizing a doppler radar to monitor the chest movement during respiration. Specifically, machine learning approach in conjunction with the radar sensor is utilized to capture the radar reflection pulse signal and its movement patterns. By analyzing the evolution of the reflected pulse while breathing, the respiratory rate can be accurately measured. In addition, the change of respiration ratio and respiration patterns can be characterized.

High Occupancy Vehicle (HOV) and High Occupancy Tolling (HOT) lanes have been commonly practiced in several jurisdictions to reduce traffic congestion and promote car pooling. Camera-based methods have been recently proposed for a cost-efficient, safe and effective HOV/HOT lane enforcement with the prevalence of video cameras in transportation imaging applications. An important step in automated lane enforcement systems is classification of localized window/windshield images to distinguish passenger from no-passenger vehicles to identify violators. This can be performed using deep convolutional neural networks (CNNs), which are shown to significantly outperform hand-crafted features in several classification tasks. Training/fine-tuning CNNs require a set of passenger/no-passenger images manually labeled by an operator, which requires substantial time and effort that can result in excessive operational cost and overhead. In this paper, we study adaptability of three popular CNNs (i.e., AlexNet, VGG-M, and GoogLeNet) across different domains in classifying passenger/no-passenger images as part of an automated HOV/HOT lane enforcement system. Our experiments over 40, 000 side-view vehicle images show many interesting insights for domain adaptability of these deep learning architectures.

The Xerox Ignite™ Educator Support System (henceforth referred to simply as Ignite™) is a data collection, analysis, and visualization workflow and software solution to assist K-12 educators. To illustrate, suppose a third-grade teacher wants to know how well her class has grasped a lesson on fractions. She would first scan her students' homework and/or exams into the Ignite system via a range of multifunctional input devices. Xerox Ignite™ reads, interprets, and analyzes the students' work in minutes. Then the teacher can select how to view the data by choosing from numerous reports. Examples are; an "at a glance" class summary that shows who needs extra help in what areas and who is ready to move on; a "context" report showing how each skill for each student is progressing over time; a grade-level performance report that helps third-grade teachers share best practices and cluster students into learning groups; and a student feedback report that tells each student what he/she needs to improve upon. Ignite™ intent is also to make it easier for districts to administer, score and evaluate content based on academic goals set for schools and students. The scanning and 'mark lifting' technology embedded into Ignite™ reduces the time needed to correct papers and frees time for the teacher to apply detailed insights to their day-to-day instruction tasks. Critical to this function is the automated reading of student marks, including handwriting, to enable the digitization of student performance at a detailed level. In this paper we present a system level description of the Ignite™ handwriting recognition module and describe the challenges and opportunities presented in an educational environment.

Keyword spotting aims to retrieve all instances of a given keyword from a document in any language. In this paper, we propose a novel script independent line based word spotting framework for offline handwritten documents based on Hidden Markov Models. The methodology simulates the keywords in model space as a sequence of character models and uses the filler models for better representation of background or non-keyword text. We propose a two stage spotting framework where the candidate keywords are further pruned using the character based background and lexicon based background model. The system deals with large vocabulary without the need for word or character segmentation. The system has been evaluated on many public dataset from several languages such as IAM for English, AMA for Arabic and LAW for Devanagari. The system outperforms the modern line based approach on the English, Arabic and Devanagari Datasets.

In this paper, we present a method for lexicon size reduction which can be used as an important pre-processing for an off-line Arabic word recognition. The method involves extraction of the dot descriptors and PAWs (Piece of Arabic Word ). Then the number and position of dots and the number of the PAWs are used to eliminate unlikely candidates. The extraction of the dot descriptors is based on defined rules followed by a convolutional neural network for verification. The reduction algorithm makes use of the combination of two features with a dynamic matching scheme. On IFN/ENIT database of 26459 Arabic handwritten word images we achieved a reduction rate of 87% with accuracy above 93%.

We propose a new algorithm for segmentation of off-line handwritten Arabic words. The algorithm segments the connected letters to smaller segments each of which contains no more than three letters. Each letter may be segmented to at most five pieces. In addition to improving the recognition of Arabic words, another potential application of the proposed segmentation method is to build lexicon of small size, consisting of no more than three letter combinations. Generally, it is very hard to generate lexicon for recognition of unconstraint handwritten Arabic documents due to the large number of words of Arabic language.The algorithm has been tested on over 6300 words from 45 different documents written by 18 writers. The system is able to segment more than 93% of the words into segments, each containing at most one letter, 6% of the words into segments that contains two letters and 3% of the words into segments that contains three letters.