Finer Resolution Levels Research Articles

SV-FPN: Small Object Feature Enhancement and Variance-Guided RoI Fusion for Feature Pyramid Networks

Small object detection is one of the research difficulties in object detection, and Feature Pyramid Networks (FPN) is a common feature extractor in deep learning; thus, improving the results of small object detection based on FPN is of great significance in this field. In this paper, SV-FPN is proposed for a small object detection task, which consists of Small Object Feature Enhancement (SOFE) and Variance-guided Region of Interest Fusion (VRoIF). When using FPN as a feature extractor, an SOFE module is designed to enhance the finer-resolution level feature maps from which the small object features are extracted. VRoIF takes the variance of RoI features as the data driver to learn the completeness of several RoI features from different feature layers, which avoids wasting information and introducing noise. Ablation experiments on three public datasets (KITTI, PASCAL VOC 07+12 and MS COCO 2017) demonstrate the effectiveness of SV-FPN, and the mean Average Precision (mAP) of SV-FPN in the three datasets achieves 41.5%, 53.9% and 38.3%, respectively.

Prediction of melter cold‐cap topology from plenum temperatures with computational fluid dynamics and machine learning

AbstractA computational fluid dynamics (CFD) model of a pilot‐scale waste‐glass melter was used to generate input data for several different machine‐learning models to predict the cold‐cap coverage from plenum temperatures. This methodology could serve as useful to provide nonvisual feedback for operational control. The machine‐learning models tested include an artificial neural network (ANN), a convolutional neural network (CNN), a random forest (RF) algorithm, and a support vector machine (SVM) method. CFD simulations with randomized cold‐cap coverage were run to generate plenum‐temperature distributions. The topology of the cold cap was predicted with each method using an input layer of selected temperature locations. The ANN was used to predict the cold‐cap coverage percentage with an accuracy of 1.2%. The accuracy of the various machine‐learning algorithms was dependent on the filter resolution employed. When using a low resolution (16‐cm filter), the CNN and ANN methods produced the best accuracy, with errors of 6.98% and 6.58%, respectively. At finer levels of resolution (4‐cm filter), the RF method produced the best accuracy, with an error of 8.12%. The ANN and SVM methods required less computational time to train than either the CNN or RF methods.

The Use of Artificial Neural Networks to Forecast the Behavior of Agent-Based Models of Pathophysiology: An Example Utilizing an Agent-Based Model of Sepsis.

Introduction: Disease states are being characterized at finer and finer levels of resolution via biomarker or gene expression profiles, while at the same time. Machine learning (ML) is increasingly used to analyze and potentially classify or predict the behavior of biological systems based on such characterization. As ML applications are extremely data-intensive, given the relative sparsity of biomedical data sets ML training of artificial neural networks (ANNs) often require the use of synthetic training data. Agent-based models (ABMs) that incorporate known biological mechanisms and their associated stochastic properties are a potential means of generating synthetic data. Herein we present an example of ML used to train an artificial neural network (ANN) as a surrogate system used to predict the time evolution of an ABM focusing on the clinical condition of sepsis.Methods: The disease trajectories for clinical sepsis, in terms of temporal cytokine and phenotypic dynamics, can be interpreted as a random dynamical system. The Innate Immune Response Agent-based Model (IIRABM) is a well-established model that utilizes known cellular and molecular rules to simulate disease trajectories corresponding to clinical sepsis. We have utilized two distinct neural network architectures, Long Short-Term Memory and Multi-Layer Perceptron, to take a time sequence of five measurements of eleven IIRABM simulated serum cytokine concentrations as input and to return both the future cytokine trajectories as well as an aggregate metric representing the patient’s state of health.Results: The ANNs predicted model trajectories with the expected amount of error, due to stochasticity in the simulation, and recognizing that the mapping from a specific cytokine profile to a state-of-health is not unique. The Multi-Layer Perceptron neural network, generated predictions with a more accurate forecasted trajectory cone.Discussion: This work serves as a proof-of-concept for the use of ANNs to predict disease progression in sepsis as represented by an ABM. The findings demonstrate that multicellular systems with intrinsic stochasticity can be approximated with an ANN, but that forecasting a specific trajectory of the system requires sequential updating of the system state to provide a rolling forecast horizon.

Single-Image Deraining via Recurrent Residual Multiscale Networks.

Existing deraining approaches represent rain streaks with different rain layers and then separate the layers from the background image. However, because of the complexity of real-world rain, such as various densities, shapes, and directions of rain streaks, it is very difficult to decompose a rain image into clean background and rain layers. In this article, we develop a novel single-image deraining method based on residual multiscale pyramid to mitigate the difficulty of rain image decomposition. To be specific, we progressively remove rain streaks in a coarse-to-fine fashion, where heavy rain is first removed in coarse-resolution levels and then light rain is eliminated in fine-resolution levels. Furthermore, based on the observation that residuals between a restored image and its corresponding rain image give critical clues of rain streaks, we regard the residuals as an attention map to remove rains in the consecutive finer level image. To achieve a powerful yet compact deraining framework, we construct our network by recurrent layers and remove rain with the same network in different pyramid levels. In addition, we design a multiscale kernel selection network (MSKSN) to facilitate our single network to remove rain streaks at different levels. In this manner, we reduce 81% of the model parameters without decreasing deraining performance compared with our prior work. Extensive experimental results on widely used benchmarks show that our approach achieves superior deraining performance compared with the state of the art.

Terminology of polymers in advanced lithography (IUPAC Recommendations 2020)

Abstract As increasingly smaller molecular materials and material structures are devised or developed for technological applications, the demands on the processes of lithography now routinely include feature sizes that are of the order of 10 nm. In reaching such a fine level of resolution, the methods of lithography have increased markedly in sophistication and brought into play 2terminology that is unfamiliar, on the one hand, to scientists tasked with the development of new lithographic materials or, on the other, to the engineers who design and operate the complex equipment that is required in modern-day processing. Publications produced by scientists need to be understood by engineers and vice versa, and these commonly arise from collaborative research that draws heavily on the terminology of two or more of the traditional disciplines. It is developments in polymer science and material science that lead progress in areas that cross traditional boundaries, such as microlithography. This document provides the exact definitions of a selection of unfamiliar terms that researchers and practitioners from different disciplines might encounter.

Estimation of spatial econometric linear models with large datasets: How big can spatial Big Data be?

Spatial econometrics is currently experiencing the Big Data revolution both in terms of the volume of data and the velocity with which they are accumulated. Regional data, employed traditionally in spatial econometric modeling, can be very large, with information that are increasingly available at a very fine resolution level such as census tracts, local markets, town blocks, regular grids or other small partitions of the territory. When dealing with spatial microeconometric models referred to the granular observations of the single economic agent, the number of observations available can be a lot higher. This paper reports the results of a systematic simulation study on the limits of the current methodologies when estimating spatial models with large datasets. In our study we simulate a Spatial Lag Model (SLM), we estimate it using Maximum Likelihood (ML), Two Stages Least Squares (2SLS) and Bayesian estimator (B), and we test their performances for different sample sizes and different levels of sparsity of the weight matrices. We considered three performance indicators, namely: computing time, storage required and accuracy of the estimators. The results show that using standard computer capabilities the analysis becomes prohibitive and unreliable when the sample size is greater than 70,000 even for low levels of sparsity. This result suggests that new approaches should be introduced to analyze the big datasets that are quickly becoming the new standard in spatial econometrics.

Hybrid artificial neural network and locally weighted regression models for lane-based short-term urban traffic flow forecasting

ABSTRACTIn recent years, there has been considerable research interest in short-term traffic flow forecasting. However, forecasting models offering a high accuracy at a fine temporal resolution (e.g. 1 or 5 min) and lane level are still rare. In this study, a combination of genetic algorithm, neural network and locally weighted regression is used to achieve optimal prediction under various input and traffic settings. The genetically optimized artificial neural network (GA-ANN) and locally weighted regression (GA-LWR) models are developed and tested, with the former forecasting traffic flow every 5-min within a 30-min period and the latter for forecasting traffic flow of a particular 5-min period of each for four lanes of an urban arterial road in Beijing, China. In particular, for morning peak and off-peak traffic flow prediction, the GA-ANN 5-min traffic flow model results in average errors of 3–5% and most 95th percentile errors of 7–14% for each of the four lanes; for the peak and off-peak time traffic flow predictions, the GA-LWR 5-min traffic flow model results in average errors of 2–4% and most 95th percentile errors are lower than 10% for each of the four lanes. When compared to previous models that usually offer average errors greater than 6–15%, such empirical findings should be of interest to and instrumental for transportation authorities to incorporate in their city- or state-wide Advanced Traveller Information Systems (ATIS).

Environmental Influences on Health Outcomes-Integrating Real-Time Exposure Measures from Homes into Longitudinal Cohort Studies: Lessons from the Field

In North America, adults and children of all ages spend much of their time indoors with the home where a majority of time is passed. Exposure to respirable particulate matter (PM) of varying sizes has been linked to cardiovascular and respiratory problems. Exposures to PM2.5 contribute to childhood respiratory ailments such as asthma. Utah collaborators with pediatric cohorts aligned with the National Institutes of Health (NIH) ECHO program and NIH PRISMS program conducted pilot projects aimed at using innovative, real-time, low-cost air quality sensors to provide spatio-temporal records of particulate matter (PM) exposures. The projects evaluated participant acceptance, feasibility, and efficacy in building exposure profiles using instrumentation, extant data, models and health information integrated with a single informatics platform. The PM sensors are small and feature secure Wi-Fi and data storage capability. More than 30 participant homes had one outdoor and two indoor sensors capturing time stamped PM counts every minute which were aggregated to daily data for initial analyses. Participants were supportive of using their personal Wi-Fi for sensor data transmission. From the PRISMS project (n=6), mixed modeling showed the daily estimated marginal mean level of small (PM2.5) and large (PM10) particle counts was lower indoors than outdoors (p < 0.001). Among 6 pediatric asthmatic participants reporting daily asthma treatment usage via eAsthma Tracker, indoor PM counts were not related to treatment usage (p > 0.05). Higher levels of PM2.5 outdoors were related to increases in asthma treatment (p < 0.001) while outdoor PM10 were not related to treatment usage (p = 0.52) in this small sample. These pilot projects demonstrated the feasibility of gathering environmental exposure data from low-cost sensors at a very fine level of resolution, and ability to integrate and analyze data across disparate data sources using the Utah PRISMS developed infrastructure.

Group-based hierarchical adaptive traffic-signal control part I: Formulation

A group-based adaptive traffic-control method for isolated signalized junctions is developed that includes a hierarchical structure comprising tactical and local levels of signal timing optimization. The control method optimizes the signal timings in adaptive traffic-control systems, and takes full advantage of flexible new technologies to incorporate the most up-to-date traffic information, as collected in real time. The definitions, combinations, and sequencing of the cycle structure stages are generated automatically using a procedure for optimizing the signal-timing plans in response to online data from traffic detectors. This new method provides a wider search space and improves the efficiency of the signal-control systems, thus improving the junction performance, minimizing delays, and maximizing capacity in real time. A multi-resolution strategy is proposed for updating the elements of the signal plans cycle-by-cycle and adjusting the current green signal timing second-by-second. The group-based variables and parameters for the proactive global-optimization method utilize lane-based predictive traffic-flow information, such as arrival and discharge rates, expressed as the slopes of polygonal delay formulas. Therefore, there is a high degree of flexibility in the tactical identification of the optimal signal plan in response to the real-time predicted traffic information, the objective function of the polygonal delay formula, and the direct differential equations for the adaptive group-based variables. The reactive local signal-control policy, which is formed based on the max-pressure strategy, is developed to locally adjust the current green signal time and to accommodate delicate demand fluctuations second-by-second at the fine-resolution level. The most appropriate cycle-structure for the tactical level of control is identified using a group-based global-optimization procedure that takes advantage of the latest available information. In part II of this study (Lee et al. 2017), the effectiveness of the proposed methods is validated based on the actualized mathematical frameworks, computer simulations, and a case study, using the appropriate computer programs.

Methods to identify the prey of invertebrate predators in terrestrial field studies.

Predation is an interaction during which an organism kills and feeds on another organism. Past and current interest in studying predation in terrestrial habitats has yielded a number of methods to assess invertebrate predation events in terrestrial ecosystems. We provide a decision tree to select appropriate methods for individual studies. For each method, we then present a short introduction, key examples for applications, advantages and disadvantages, and an outlook to future refinements. Video and, to a lesser extent, live observations are recommended in studies that address behavioral aspects of predator–prey interactions or focus on per capita predation rates. Cage studies are only appropriate for small predator species, but often suffer from a bias via cage effects. The use of prey baits or analyses of prey remains are cheaper than other methods and have the potential to provide per capita predation estimates. These advantages often come at the cost of low taxonomic specificity. Molecular methods provide reliable estimates at a fine level of taxonomic resolution and are free of observer bias for predator species of any size. However, the current PCR‐based methods lack the ability to estimate predation rates for individual predators and are more expensive than other methods. Molecular and stable isotope analyses are best suited to address systems that include a range of predator and prey species. Our review of methods strongly suggests that while in many cases individual methods are sufficient to study specific questions, combinations of methods hold a high potential to provide more holistic insights into predation events. This review presents an overview of methods to researchers that are new to the field or to particular aspects of predation ecology and provides recommendations toward the subset of suitable methods to identify the prey of invertebrate predators in terrestrial field research.

NAD9/NAD7 (mitochondrial nicotinamide adenine dinucleotide dehydrogenase gene)—A new “Holy Grail” phylogenetic and DNA-barcoding marker for Arcellinida (Amoebozoa)?

Molecular phylogeny is an indispensable tool for assessing evolutionary relationships among protists. The most commonly used marker is the small subunit ribosomal RNA gene, a conserved gene present in many copies in the nuclear genomes. However, this marker is not variable enough at a fine-level taxonomic scale, and intra-genomic polymorphism has already been reported. Finding a marker that could be useful at both deep and fine taxonomic resolution levels seemed like a utopic dream. We designed Amoebozoa-specific primers to amplify a region including partial sequences of two subunits of the mitochondrial nicotinamide adenine dinucleotide dehydrogenase gene (NAD9/NAD7). We applied them to arcellinids belonging to distantly related genera (Arcella, Difflugia, Netzelia and Hyalosphenia) and to Arcellinid-rich environmental samples to obtain additional Amoebozoa sequences. Tree topology was congruent with previous phylogenies, all nodes being highly supported, suggesting that this marker is well-suited for deep phylogenies in Arcellinida and perhaps Amoebozoa. Furthermore, it enabled discrimination of close-related taxa. This short genetic marker (ca. 250bp) can therefore be used at different taxonomic levels, due to a fast-varying intergenic region presenting either a small intergenic sequence or an overlap, depending on the species.

A New Cascade Model for the Hierarchical Joint Classification of Multitemporal and Multiresolution Remote Sensing Data

In this paper, we propose a novel method for the joint classification of both multidate and multiresolution remote sensing imagery, which represents an important and relatively unexplored classification problem. The proposed classifier is based on an explicit hierarchical graph-based model that is sufficiently flexible to address a coregistered time series of images collected at different spatial resolutions. Within this framework, a novel element of the proposed approach is the use of multiple quadtrees in cascade, each associated with the images available at each observation date in the considered time series. For each date, the input images are inserted in a hierarchical structure on the basis of their resolutions, whereas missing levels are filled in with wavelet transforms of the images embedded in finer-resolution levels. This approach is aimed at both exploiting multiscale information, which is known to play a crucial role in high-resolution image analysis, and supporting input images acquired at different resolutions in the input time series. The experimental results are shown for multitemporal and multiresolution optical data.

Adaptive areal elimination (AAE): A transparent way of disclosing protected spatial datasets

Geographical masking is the conventional solution to protect the privacy of individuals involved in confidential spatial point datasets. The masking process displaces confidential locations to protect individual privacy while maintaining a fine level of spatial resolution. The adaptive form of this process aims to further minimize the displacement error by taking into account the underlying population density. We describe an alternative adaptive geomasking method, referred to as Adaptive Areal Elimination (AAE). AAE creates areas of a minimum K-anonymity and then original points are either randomly perturbed within the areas or aggregated to the median centers of the areas. In addition to the masked points, K-anonymized areas can be safely disclosed as well without increasing the risk of re-identification. Using a burglary dataset from Vienna, AAE is compared with an existing adaptive geographical mask, the donut mask. The masking methods are evaluated for preserving a predefined K-anonymity and the spatial characteristics of the original points. The spatial characteristics are assessed with four measures of spatial error: displaced distance, correlation coefficient of density surfaces, hotspots' divergence, and clusters' specificity. Masked points from point aggregation of AAE have the highest spatial error in all the measures but the displaced distance. In contrast, masked points from the donut mask are displaced the least, preserve the original spatial clusters better, have the highest clusters' specificity and correlation coefficient of density surfaces. However, when the donut mask is adapted to achieve an actual K-anonymity, the random perturbation of AAE introduces less spatial error than the donut mask for all the measures of spatial error.

Optimization of the prediction of second refined wavelet coefficients in electron structure calculations

Abstract In wavelet-based solution of eigenvalue-type differential equations, like the Schrödinger equation, refinement in the resolution of the solution is a costly task, as the number of the potential coefficients in the wavelet expansion of the solution increases exponentially with the resolution. Predicting the magnitude of the next resolution level coefficients from an already existing solution in an economic way helps to either refine the solution,or to select the coefficients, which are to be included into the next resolution level calculations, or to estimate the magnitude of the error of the solution. However, after accepting a solution with a predicted refinement as a basis, the error can still be estimated by a second prediction, i.e., from a prediction to the second finer resolution level coefficients. These secondary predicted coefficients are proven to be oscillating around the values of the wavelet expansion coefficients of the exact solution. The optimal averaging of these coefficients is presented in the following paper using a sliding average with three optimized coefficients for simple, one-dimensional electron structures.

Response to Tucker & Vuchinich (2015): Behavioral economics in the broader context of addiction science.

In their commentary on our recent study 1, Tucker & Vuchinich 2 argue that the core feature of behavioral economics (BE) is the analysis of temporally protracted molar patterns of choice behavior. We appreciate this view, and agree that BE has been highly effective in characterizing patterns of substance use behavior over time. However, we contend that BE is also well-suited for examining behavior at a finer level of resolution, investigating moment-to-moment influences that contribute to decisions regarding substance use and related behaviors. To use Tucker & Vuchinich's 2 example, John may not be drinking alcohol right now, but his decision to do so (or not) stems from a dynamic interplay between stable individual difference factors (e.g. personality traits, attitudes) and situation-specific, within-person influences (e.g. craving, negative affect) that serve as immediate precursors of behavior. These proximal influences are viewed increasingly as critical contributors to substance misuse 3, 4. In our view, the argument that momentary dynamic changes, whether in cognitive states or BE indices, can be ‘safely ignored’ 2 runs counter to current research in addiction science and psychological science more broadly 5. Ignoring such within-person variation closes the door on a potentially fruitful direction for BE 6, 7. Considerable evidence suggests that the BE indices of demand and delay discounting exhibit meaningful change in response to relevant manipulations. Delay discounting rates, for instance, change in response to drug deprivation, environmental cues and acute drug administration (for a review, see 7). The level of alcohol demand is enhanced dynamically by many of the same manipulations 8, 9. Furthermore, demand is also sensitive to pharmacological interventions 10 and contextual factors such as next-day responsibilities 11. Although this is still a nascent research area, these promising findings are consistent with an efficient cause interpretation. This may help to clarify situational factors that impact choice behavior in addiction and, in turn, provide a point of entry for in-the-moment interventions 12 and the development of personalized medicine applications 13. Tucker & Vuchinich 2 noted that only the molar approach to BE has proved useful in extensions to drinking in the natural environment. Whether this holds for laboratory manipulations of demand remains an empirical question. Recent research has laid the necessary groundwork for examining how changes in BE indices translate into drinking behavior, including linking changes in demand with actual consumption in the laboratory 14-16. Our study was designed to provide proof of concept for testing whether within-person changes in demand predict drinking in the laboratory and the natural environment, as has been demonstrated for within-person variation in craving 17, 18. We agree with Tucker & Vuchinich that both efficient and final cause interpretations of BE indices are legitimate 2, as they have also discussed elsewhere (e.g. 19). Ultimately, it is worth emphasizing that BE was developed from integrating methods across disciplines, and addiction science relies increasingly upon integration across multiple domains and levels of analyses. We would argue against an interpretation of BE that impedes its incorporation into a broader range of addiction research. None. Funding for this work was provided by National Institute on Alcohol Abuse and Alcoholism Grants R01 AA 019546 to D.M.McC. and T32 AA 013526 to K.J.S.

3D Finger CAPE: Clicking Action and Position Estimation under Self-Occlusions in Egocentric Viewpoint.

In this paper we present a novel framework for simultaneous detection of click action and estimation of occluded fingertip positions from egocentric viewed single-depth image sequences. For the detection and estimation, a novel probabilistic inference based on knowledge priors of clicking motion and clicked position is presented. Based on the detection and estimation results, we were able to achieve a fine resolution level of a bare hand-based interaction with virtual objects in egocentric viewpoint. Our contributions include: (i) a rotation and translation invariant finger clicking action and position estimation using the combination of 2D image-based fingertip detection with 3D hand posture estimation in egocentric viewpoint. (ii) a novel spatio-temporal random forest, which performs the detection and estimation efficiently in a single framework. We also present (iii) a selection process utilizing the proposed clicking action detection and position estimation in an arm reachable AR/VR space, which does not require any additional device. Experimental results show that the proposed method delivers promising performance under frequent self-occlusions in the process of selecting objects in AR/VR space whilst wearing an egocentric-depth camera-attached HMD.

A Novel Semi-Supervised Method for Obtaining Finer Resolution Urban Extents Exploiting Coarser Resolution Maps

In this work, we present a new semi-supervised strategy for obtaining finer spatial resolution urban maps from coarser resolution satellite data. Our method first uses a coarse resolution map as a source of training data. Then, we use semi-supervised learning in order to refine the set of initial (labeled) training samples by the inclusion of additional (reliable) unlabeled samples at the finer resolution level, in fully automatic fashion. The new unlabeled samples are automatically generated by our proposed methodology, which only requires a limited number of initial labeled samples for initialization purposes. Then, we conduct land cover classification (at the finer spatial resolution level) using a probabilistic multinomial logistic regression (MLR) classifier-in both supervised and semi-supervised fashion-by considering different numbers of labeled and unlabeled samples. In order to exploit spatial information, we use a Markov random field (MRF)-based postprocessing strategy to refine the obtained classification results. In order to test our concept, we use a global dataset: the European Space Agency's GlobCover product, as the coarser resolution map (300-m spatial resolution). Our experimental evaluation is further conducted using Landsat data (30-m spatial resolution) collected over three different locations in the city of Sao Paulo, Brazil, and over two different locations in the city of Guangzhou, China. We obtain promising results in the generation of finer resolution urban extent maps using very limited training samples, derived in all cases from the GlobCover product. These experiments suggest the potential of GlobCover to provide reliable training data in order to support mapping of urban areas at a global scale.

A model for work activity schedules with synchronization for multiple-worker households

This paper presents a hybrid discrete choice-duration model for work activity scheduling with interactions between workers in a multiple-worker household. The model operates in discrete space with a fine level of temporal resolution. The key innovative feature of the model is the introduction of intra-household interactions through worker schedule synchronization mechanisms. The model was estimated based on a large Household Travel Survey from the San Francisco Bay Area. The estimation results confirmed that individual work schedules for workers in a multiple-worker household are subject to strong synchronization and should be modelled jointly rather than independently. In particular, workers in the same household tend to align their schedules and create time window overlaps for joint activities before and after work. Relative strength of the synchronization mechanisms proved to be a function of the person characteristics and household composition.

Co-word based thematic analysis of renewable energy (1990–2010)

This article describes an analysis of keywords which was aimed at revealing publication patterns in the field of renewable energy, including the temporal evolution of its different research lines over the last two decades. To this end, we first retrieved the records of the sample, then we processed the keywords to resolve their obvious problems of synonymy and to limit the study to those most used. The final results showed a clear increase in scientific production related to alternative energies, and a structure corresponding to five major clusters which, at a finer level of resolution, were decomposed into 22. We analyzed the structure of the clusters and their temporal evolution, paying particular attention to uncovering the bursty periods of the different lines of research.

Damage detection on a full-scale highway sign structure with a distributed wireless sensor network

Wireless sensor networks (WSNs) have emerged as a novel solution to many of the challenges of structural health monitoring (SHM) in civil engineering structures. While research projects using WSNs are ongoing worldwide, implementations of WSNs on full-scale structures are limited. In this study, a WSN is deployed on a full-scale 17.3m-long, 11-bay highway sign support structure to investigate the ability to use vibration response data to detect damage induced in the structure. A multi-level damage detection strategy is employed for this structure: the Angle-between-String-and-Horizon (ASH) flexibility-based algorithm as the Level I and the Axial Strain (AS) flexibility-based algorithm as the Level II. For the proposed multi-level damage detection strategy, a coarse resolution Level I damage detection will be conducted first to detect the damaged region(s). Subsequently, a fine resolution Level II damage detection will be conducted in the damaged region(s) to locate the damaged element(s). Several damage cases are created on the full-scale highway sign support structure to validate the multi-level detection strategy. The multi-level damage detection strategy is shown to be successful in detecting damage in the structure in these cases.