Search

The United States (US) apple industry relies heavily on semi‐skilled manual labor force for essential field operations such as training, pruning, blossom and green fruitlet thinning, and harvesting. Blossom thinning is one of the crucial crop‐load management practices to achieve desired crop load, fruit quality, and return bloom. While several techniques such as chemical and mechanical thinning are available for large‐scale blossom thinning, such approaches often yield unpredictable thinning results and may damage the canopy, spurs, and leaf tissue. Hence, growers still depend on laborious, labor‐intensive, and expensive manual hand blossom thinning for desired thinning outcomes. This research presents a robotic solution for precision blossom thinning in apple orchards using a deep learning‐based computer vision system, a six‐degrees‐of‐freedom UR5e robotic manipulator, and an electrically actuated miniature end‐effector. The integrated robotic system was evaluated in a commercial apple orchard which showed promising results for targeted and selective blossom thinning. Two thinning approaches, center and boundary thinning, were investigated to evaluate the system's ability to remove varying proportions of flowers from apple flower clusters. During boundary thinning, the end‐effector was actuated around the cluster boundary, while center thinning involved end‐effector actuation only at the cluster centroid for a fixed duration of 2 s. Field evaluation results showed that the boundary thinning approach thinned 67.2% of flowers from the targeted clusters with a cycle time of 9.0 s per cluster, whereas the center thinning approach thinned 59.4% of flowers with a cycle time of 7.2 s per cluster. Upon further improvement for commercial adoption, the proposed system could help address problems faced by apple growers with current hand, chemical, and mechanical blossom thinning approaches. [ABSTRACT FROM AUTHOR]

The ecology of forest ecosystems depends on the composition of trees. Capturing fine-grained information on individual trees at broad scales provides a unique perspective on forest ecosystems, forest restoration, and responses to disturbance. Individual tree data at wide extents promises to increase the scale of forest analysis, biogeographic research, and ecosystem monitoring without losing details on individual species composition and abundance. Computer vision using deep neural networks can convert raw sensor data into predictions of individual canopy tree species through labeled data collected by field researchers. Using over 40,000 individual tree stems as training data, we create landscape-level species predictions for over 100 million individual trees across 24 sites in the National Ecological Observatory Network (NEON). Using hierarchical multi-temporal models fine-tuned for each geographic area, we produce open-source data available as 1 km2 shapefiles with individual tree species prediction, as well as crown location, crown area, and height of 81 canopy tree species. Site-specific models had an average performance of 79% accuracy covering an average of 6 species per site, ranging from 3 to 15 species per site. All predictions are openly archived and have been uploaded to Google Earth Engine to benefit the ecology community and overlay with other remote sensing assets. We outline the potential utility and limitations of these data in ecology and computer vision research, as well as strategies for improving predictions using targeted data sampling. Detailed maps of tree species can be used for forest analysis, biogeographic research and ecosystem monitoring. This study provides individual tree species maps for 24 sites in the United States National Ecological Observatory Network (NEON), covering 80 tree species and 100 million individuals. [ABSTRACT FROM AUTHOR]

Simple Summary: Breast cancer is a significant cause of female cancer-related deaths in the US. Checking how severe the cancer is helps in planning treatment. Modern AI methods are good at grading cancer, but they are not used much in hospitals yet. We developed and utilized ensemble deep learning algorithms for addressing the tasks of classifying (1) breast cancer subtype and (2) breast cancer invasiveness from whole slide image (WSI) histopathology slides. The ensemble models used were based on convolutional neural networks (CNNs) known for extracting distinctive features crucial for accurate classification. In this paper, we provide a comprehensive analysis of these models and the used methodology for breast cancer diagnosis tasks. Cancer diagnosis and classification are pivotal for effective patient management and treatment planning. In this study, a comprehensive approach is presented utilizing ensemble deep learning techniques to analyze breast cancer histopathology images. Our datasets were based on two widely employed datasets from different centers for two different tasks: BACH and BreakHis. Within the BACH dataset, a proposed ensemble strategy was employed, incorporating VGG16 and ResNet50 architectures to achieve precise classification of breast cancer histopathology images. Introducing a novel image patching technique to preprocess a high-resolution image facilitated a focused analysis of localized regions of interest. The annotated BACH dataset encompassed 400 WSIs across four distinct classes: Normal, Benign, In Situ Carcinoma, and Invasive Carcinoma. In addition, the proposed ensemble was used on the BreakHis dataset, utilizing VGG16, ResNet34, and ResNet50 models to classify microscopic images into eight distinct categories (four benign and four malignant). For both datasets, a five-fold cross-validation approach was employed for rigorous training and testing. Preliminary experimental results indicated a patch classification accuracy of 95.31% (for the BACH dataset) and WSI image classification accuracy of 98.43% (BreakHis). This research significantly contributes to ongoing endeavors in harnessing artificial intelligence to advance breast cancer diagnosis, potentially fostering improved patient outcomes and alleviating healthcare burdens. [ABSTRACT FROM AUTHOR]

A significant percentage of bridges in the United States are serving beyond their 50-year design life, and many of them are in poor condition, making them vulnerable to fatigue cracks that can result in catastrophic failure. However, current fatigue crack inspection practice based on human vision is time-consuming, labor intensive, and prone to error. We present a novel human-centered bridge inspection methodology to enhance the efficiency and accuracy of fatigue crack detection by employing advanced technologies including computer vision and augmented reality (AR). In particular, a computer vision-based algorithm is developed to enable near-real-time fatigue crack detection by analyzing structural surface motion in a short video recorded by a moving camera of the AR headset. The approach monitors structural surfaces by tracking feature points and measuring variations in distances between feature point pairs to recognize the motion pattern associated with the crack opening and closing. Measuring distance changes between feature points, as opposed to their displacement changes before this improvement, eliminates the need of camera motion compensation and enables reliable and computationally efficient fatigue crack detection using the nonstationary AR headset. In addition, an AR environment is created and integrated with the computer vision algorithm. The crack detection results are transmitted to the AR headset worn by the bridge inspector, where they are converted into holograms and anchored on the bridge surface in the 3D real-world environment. The AR environment also provides virtual menus to support human-in-the-loop decision-making to determine optimal crack detection parameters. This human-centered approach with improved visualization and human–machine collaboration aids the inspector in making well-informed decisions in the field in a near-real-time fashion. The proposed crack detection method is comprehensively assessed using two laboratory test setups for both in-plane and out-of-plane fatigue cracks. Finally, using the integrated AR environment, a human-centered bridge inspection is conducted to demonstrate the efficacy and potential of the proposed methodology. [ABSTRACT FROM AUTHOR]

Outlier detection plays a critical role in building operation optimization and data quality maintenance. However, existing methods often struggle with the complexity and variability of building energy data, leading to poorly generalized and explainable results. To address the gap, this study introduces a novel Vision-based Outlier Detection (VOD) approach, leveraging computer vision models to spot outliers in the building energy records. The models are trained to identify outliers by analyzing the load shapes in 2D time series plots derived from the energy data. The VOD approach is tested on four years of workday time-series electricity consumption data from 290 commercial buildings in the United States. Two distinct models are developed for different usage purposes, namely a classification model for broad-level outlier detection and an object detection model for the demands of precise pinpointing of outliers. The classification model is also interpreted via Grad-CAM to enhance its usage reliability. The classification model achieves an F1 score of 0.88, and the object detection model achieves an Average Precision (AP) of 0.84. VOD is a very efficient path to identifying energy consumption outliers in building operations, paving the way for the enhancement of building energy data quality, operation efficiency, and energy savings. [ABSTRACT FROM AUTHOR]

A longstanding line of research in urban studies explores how cities can be understood through their appearance. However, what remains unclear is to what extent urban dwellers' everyday life can be explained by the visual clues of the urban environment. In this paper, we address this question by applying a computer vision model to 27 million street view images across 80 counties in the United States. Then, we use the spatial distribution of notable urban features identified through the street view images, such as street furniture, sidewalks, building façades, and vegetation, to predict the socioeconomic profiles of their immediate neighborhood. Our results show that these urban features alone can account for up to 83% of the variance in people's travel behavior, 62% in poverty status, 64% in crime, and 68% in health behaviors. The results outperform models based on points of interest (POI), population, and other demographic data alone. Moreover, incorporating urban features captured from street view images can improve the explanatory power of these other methods by 5% to 25%. We propose "urban visual intelligence" as a process to uncover hidden city profiles, infer, and synthesize urban information with computer vision and street view images. This study serves as a foundation for future urban research interested in this process and understanding the role of visual aspects of the city. [ABSTRACT FROM AUTHOR]

The ornamental crop industry is an important contributor to the economy in the United States. The industry has been facing challenges due to continuously increasing labor and agricultural input costs. Sensing and automation technologies have been introduced to reduce labor requirements and to ensure efficient management operations. This article reviews current sensing and automation technologies used for ornamental nursery crop production and highlights prospective technologies that can be applied for future applications. Applications of sensors, computer vision, artificial intelligence (AI), machine learning (ML), Internet-of-Things (IoT), and robotic technologies are reviewed. Some advanced technologies, including 3D cameras, enhanced deep learning models, edge computing, radio-frequency identification (RFID), and integrated robotics used for other cropping systems, are also discussed as potential prospects. This review concludes that advanced sensing, AI and robotic technologies are critically needed for the nursery crop industry. Adapting these current and future innovative technologies will benefit growers working towards sustainable ornamental nursery crop production. [ABSTRACT FROM AUTHOR]

A smart city is a city that binds together technology, society, and government to enable the existence of a smart economy, smart mobility, smart environment, smart living, smart people, and smart governance in order to reduce the environmental impact of cities and improve life quality. The first step to achieve a fully connected smart city is to start with smaller modules such as smart homes and smart buildings with energy management systems. Buildings are responsible for a third of the total energy consumption; moreover, heating, ventilation, and air conditioning (HVAC) systems account for more than half of the residential energy consumption in the United States. Even though connected thermostats are widely available, they are not used as intended since most people do not have the expertise to control this device to reduce energy consumption. It is commonly set according to their thermal comfort needs; therefore, unnecessary energy consumption is often caused by wasteful behaviors and the estimated energy saving is not reached. Most studies in the thermal comfort domain to date have relied on simple activity diaries to estimate metabolic rate and fixed values of clothing parameters for strategies to set the connected thermostat's setpoints because of the difficulty in tracking those variables. Therefore, this paper proposes a strategy to save energy by dynamically changing the setpoint of a connected thermostat by human activity recognition based on computer vision preserving the occupant's thermal comfort. With the use of a depth sensor in conjunction with an RGB (Red–Green–Blue) camera, a methodology is proposed to eliminate the most common challenges in computer vision: background clutter, partial occlusion, changes in scale, viewpoint, lighting, and appearance on human detection. Moreover, a Recurrent Neural Network (RNN) is implemented for human activity recognition (HAR) because of its data's sequential characteristics, in combination with physiological parameters identification to estimate a dynamic metabolic rate. Finally, a strategy for dynamic setpoints based on the metabolic rate, predicted mean vote (PMV) parameter and the air temperature is simulated using EnergyPlus™ to evaluate the energy consumption in comparison with the expected energy consumption with fixed value setpoints. This work contributes with a strategy to reduce energy consumption up to 15% in buildings with connected thermostats from the successful implementation of the proposed method. [ABSTRACT FROM AUTHOR]

Annotation and analysis of sports videos is a time-consuming task that, once automated, will provide benefits to coaches, players, and spectators. American football, as the most watched sport in the United States, could especially benefit from this automation. Manual annotation and analysis of recorded videos of American football games is an inefficient and tedious process. Currently, most college football programs focus on annotating offensive formations to help them develop game plans for their upcoming games. As a first step to further research for this unique application, we use computer vision and deep learning to analyze an overhead image of a football play immediately before the play begins. This analysis consists of locating individual football players and labeling their position or roles, as well as identifying the formation of the offensive team. We obtain greater than 90% accuracy on both player detection and labeling, and 84.8% accuracy on formation identification. These results prove the feasibility of building a complete American football strategy analysis system using artificial intelligence. Collecting a larger dataset in real-world situations will enable further improvements. This would likewise enable American football teams to analyze game footage quickly. [ABSTRACT FROM AUTHOR]

Modern roundabouts are popular intersection control designs in many countries and are increasingly popular in the United States. Roundabouts facilitate reduced vehicle delays with naturally optimized conflict resolution for turning traffic, which also reduces the risks of severe crashes. However, evaluating the roundabout capacity for multilane configurations can be challenging due to the randomized decision making to accept or reject a headway to enter the roundabout. In addition, considering the follow-up headway between two vehicles entering the roundabout from the same lane is critical to evaluate accurate roundabout capacity. Several manual techniques are popularly used to evaluate roundabout capacity using computer vision powered by multiple video cameras and observers. However, manual processing of videos with a narrow field of view (FoV) requires significant computational effort. Traditional techniques used in manual processing involve a complex two-step time stamp recording and interpreting the parameters required for capacity evaluation. In this case study, a one-step gap-based methodology is proposed to accurately measure the roundabout capacity parameters. In addition, a computer vision algorithm is developed to integrate with deep learning to detect and track vehicular traffic in a multilane roundabout. A software-defined technique is developed to process videos with wider FoV powered by unmanned aerial vehicles (UAVs) and evaluate roundabout capacity parameters, such as accept, reject, and follow-up headways. Furthermore, the mean critical headway is calculated using a maximum likelihood estimation method. The evaluated roundabout capacity parameters are compared with manual technique results, and the corresponding values are published in the current standards. [ABSTRACT FROM AUTHOR]

Coastal hazard events such as hurricanes pose a significant threat to coastal communities. Disaster relief is essential to mitigating damage from these catastrophes; therefore, accurate and efficient damage assessment is key to evaluating the extent of damage inflicted on coastal cities and structures. Historically, this process has been carried out by human task forces that manually take post-disaster images and identify the damaged areas. While this method has been well established, current digital tools used for computer vision tasks such as artificial intelligence and machine learning put forth a more efficient and reliable method for assessing post-disaster damage. Using transfer learning on three advanced neural networks, ResNet, MobileNet, and EfficientNet, we applied techniques for damage classification and damaged object detection to our post-hurricane image dataset comprised of damaged buildings from the coastal region of the southeastern United States. Our dataset included 1000 images for the classification model with a binary classification structure containing classes of floods and non-floods and 800 images for the object detection model with four damaged object classes damaged roof, damaged wall, flood damage, and structural damage. Our damage classification model achieved 76 % overall accuracy for ResNet and 87 % overall accuracy for MobileNet. The F1 score for MobileNet was also 9 % higher than the F1 score of ResNet at 0.88. Our damaged object detection model achieved predominant predictions of the four damaged object classes, with MobileNet attaining the highest overall confidence score of 97.58 % in its predictions. The object detection results highlight the model's ability to successfully identify damaged areas of buildings and structures from images in a time span of seconds, which is necessary for more efficient damage assessment. Thus, we show that this level of accuracy for our damage assessment using artificial intelligence is akin to the accuracy of manual damage assessments while also completing the assessment in a drastically shorter time span. [ABSTRACT FROM AUTHOR]

Copyright of Apuntes Universitarios: Revista de Investigación is the property of Universidad Peruana Union Filiar Tarapoto and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Previous studies on public transit in cities found the positive role of bus shelters in promoting bus ridership. However, a large-scale and comparative investigation of bus shelter status has yet to be conducted, leaving a significant knowledge gap. To fill this gap, this research examined the socio-spatial patterns of bus shelters in 20 small- and medium-sized cities in the United States by employing a deep learning-based computer vision analysis with large-scale street-view images. The results revealed a regional difference in the bus shelter scores (range: 30.2–52.1 %). Overall, there are more bus shelters in neighborhoods with higher population densities or higher proportions of minority populations. However, there are nine cities where neighborhoods with higher proportions of minority populations are not significantly correlated with more bus shelters, suggesting issues of mobility injustice. This study is one of the first to combine an AI method with emerging urban data to examine the socio-spatial patterns of bus shelters in cities. • The socio-spatial patterns of bus shelters were investigated. • A deep learning-based computer vision analysis with street-view images was employed. • A notable regional difference in the bus shelter scores was identified. • There are more bus shelters in neighborhoods with higher population densities in 18 cities. • There are more bus shelters in neighborhoods with higher proportions of minority populations in 11 cities. [ABSTRACT FROM AUTHOR]

• A machine vision-based prototype was developed for online quality grading of sweetpotatoes. • Full surface was inspected by imaging sweetpotatoes rotating forward on a roller conveyor. • A YOLOv8-based pipeline was developed for segmenting, tracking, and grading sweetpotatoes in real-time. • The prototype achieved an overall accuracy of nearly 92 % in grading four classes of sweetpotatoes. Rapid advances in machine vision and artificial intelligence (AI) technologies have generated high interest in developing automated grading and sorting solutions for fresh produce of specialty crops. Sweetpotatoes are an economically important crop in the United States and beyond, but grading and sorting the commodity, especially for surface defects, remains a labor-intensive operation at commercial packing facilities. In addition to incurring high labor costs, manual grading is subjective and prone to human assessment error. This study was therefore aimed to develop automated grading technology based on machine vision and state-of-the-art AI to reduce labor dependence and improve quality assessment for sweetpotato packing lines. The developed grading system prototype consisted of a custom-designed roller conveyor to transport and rotate sweetpotatoes and an overhead color camera to perform full-surface quality assessment for size and surface defects. A YOLOv8-based computer vision pipeline was developed to segment and track each sweetpotato traveling on the conveyor and analyze sample quality conditions in real-time. The YOLOv8l model coupled with the BoT-SORT tracker achieved the overall Higher Order Tracking Accuracy (HOTA) score of 0.952. Based on the instance segmentation by YOLOv8l, the length and width of sweetpotatoes were estimated with percentage accuracies of 96.7 % and 90.3 %, respectively, while the instance segmentation for three classes of sweetpotatoes according to surface defect conditions yielded an overall accuracy of 82.6 %, regardless of samples. The final sample grading, which was realized by taking into account the quality information recorded from a sequence of multi-view images for each sample, achieved an overall accuracy of 91.7 % in grading four classes of sweetpotatoes. The developed prototype holds promise for next-stage integration with sorting mechanisms into a full-fledged sorting system and will be eventually beneficial for sweetpotatoes packers. [ABSTRACT FROM AUTHOR]

Objectives: Built environments can affect health, but data in many geographic areas are limited. We used a big data source to create national indicators of neighborhood quality and assess their associations with health. Methods: We leveraged computer vision and Google Street View images accessed from December 15, 2017, through July 17, 2018, to detect features of the built environment (presence of a crosswalk, non–single-family home, single-lane roads, and visible utility wires) for 2916 US counties. We used multivariate linear regression models to determine associations between features of the built environment and county-level health outcomes (prevalence of adult obesity, prevalence of diabetes, physical inactivity, frequent physical and mental distress, poor or fair self-rated health, and premature death [in years of potential life lost]). Results: Compared with counties with the least number of crosswalks, counties with the most crosswalks were associated with decreases of 1.3%, 2.7%, and 1.3% of adult obesity, physical inactivity, and fair or poor self-rated health, respectively, and 477 fewer years of potential life lost before age 75 (per 100 000 population). The presence of non–single-family homes was associated with lower levels of all health outcomes except for premature death. The presence of single-lane roads was associated with an increase in physical inactivity, frequent physical distress, and fair or poor self-rated health. Visible utility wires were associated with increases in adult obesity, diabetes, physical and mental distress, and fair or poor self-rated health. Conclusions: The use of computer vision and big data image sources makes possible national studies of the built environment's effects on health, producing data and results that may inform national and local decision-making. [ABSTRACT FROM AUTHOR]

Insect populations are changing rapidly, and monitoring these changes is essential for understanding the causes and consequences of such shifts. However, large‐scale insect identification projects are time‐consuming and expensive when done solely by human identifiers. Machine learning offers a possible solution to help collect insect data quickly and efficiently.Here, we outline a methodology for training classification models to identify pitfall trap‐collected insects from image data and then apply the method to identify ground beetles (Carabidae). All beetles were collected by the National Ecological Observatory Network (NEON), a continental scale ecological monitoring project with sites across the United States. We describe the procedures for image collection, image data extraction, data preparation, and model training, and compare the performance of five machine learning algorithms and two classification methods (hierarchical vs. single‐level) identifying ground beetles from the species to subfamily level. All models were trained using pre‐extracted feature vectors, not raw image data. Our methodology allows for data to be extracted from multiple individuals within the same image thus enhancing time efficiency, utilizes relatively simple models that allow for direct assessment of model performance, and can be performed on relatively small datasets.The best performing algorithm, linear discriminant analysis (LDA), reached an accuracy of 84.6% at the species level when naively identifying species, which was further increased to >95% when classifications were limited by known local species pools. Model performance was negatively correlated with taxonomic specificity, with the LDA model reaching an accuracy of ~99% at the subfamily level. When classifying carabid species not included in the training dataset at higher taxonomic levels species, the models performed significantly better than if classifications were made randomly. We also observed greater performance when classifications were made using the hierarchical classification method compared to the single‐level classification method at higher taxonomic levels.The general methodology outlined here serves as a proof‐of‐concept for classifying pitfall trap‐collected organisms using machine learning algorithms, and the image data extraction methodology may be used for nonmachine learning uses. We propose that integration of machine learning in large‐scale identification pipelines will increase efficiency and lead to a greater flow of insect macroecological data, with the potential to be expanded for use with other noninsect taxa. [ABSTRACT FROM AUTHOR]

Cervical cancer if detected early has an upward of 89% survival rate. The leading tool in identifying cervical cancer in its infancy is the Papanicolaou (Pap smear) test, which since its introduction dropped cervical cancer related deaths by 60%. The Pap smear test or Liquid-based Cytology (LBC) is a time-consuming procedure that requires a pathologist to manually identify cervical cells that may be in the middle of the processes that indicate cervical cancer. Unfortunately, due to the expenses related to conducting the Pap smear test many women are blocked from access to it and this leads to over 4000 women dying annually from cervical cancer in the United States alone. The aim of this research is to automate the methods used by pathologists in conducting the Pap smear or LBC. We show that using machine vision, ensemble learning and deep learning methods a significant portion of the Pap smear can be done automated. We set out to extract cells and cell clusters and classify those samples based on the Bethesda System for reporting cervical cytology. Achieving an accuracy of 90.4% and 91.6% for the ensemble learning and deep learning methods respectively when evaluated with a five-fold cross-validation demonstrates promise in the creation of an automated Pap smear screening test. [ABSTRACT FROM AUTHOR]

Premise: Apple orchards in the United States are under constant threat from a large number of pathogens and insects. Appropriate and timely deployment of disease management depends on early disease detection. Incorrect and delayed diagnosis can result in either excessive or inadequate use of chemicals, with increased production costs and increased environmental and health impacts. Methods and Results: We have manually captured 3651 high‐quality, real‐life symptom images of multiple apple foliar diseases, with variable illumination, angles, surfaces, and noise. A subset of images, expert‐annotated to create a pilot data set for apple scab, cedar apple rust, and healthy leaves, was made available to the Kaggle community for the Plant Pathology Challenge as part of the Fine‐Grained Visual Categorization (FGVC) workshop at the 2020 Computer Vision and Pattern Recognition conference (CVPR 2020). Participants were asked to use the image data set to train a machine learning model to classify disease categories and develop an algorithm for disease severity quantification. The top three area under the ROC curve (AUC) values submitted to the private leaderboard were 0.98445, 0.98182, and 0.98089. We also trained an off‐the‐shelf convolutional neural network on this data for disease classification and achieved 97% accuracy on a held‐out test set. Discussion: This data set will contribute toward development and deployment of machine learning–based automated plant disease classification algorithms to ultimately realize fast and accurate disease detection. We will continue to add images to the pilot data set for a larger, more comprehensive expert‐annotated data set for future Kaggle competitions and to explore more advanced methods for disease classification and quantification. [ABSTRACT FROM AUTHOR]

Robotic weed control for vegetables is necessary to increase crop productivity, avoid intensive hand weeding as labour shortages in developed countries such as United States has led to a surge in food production costs. However, development of a reliable, intelligent robotic system for weed control in real-time for vegetables still remains a challenging task. The main issue arises while distinguishing crops from weeds in real-time. In this paper, a novel technique to crop signalling to distinguish crops from in-row weeds in complex natural scenarios, such as high weed densities commonly found on organic farms, in real-time using machine vision is presented. Crop signalling is a simple and low-cost technique in which a signalling compound is produced by or applied to the crop and where the signalling compound is machine readable and helps to create visual features that uniquely distinguish the crops from weeds. The crop and weed mapping algorithm presented here were specially designed and developed for a vision-based weeding robot equipped with a micro-jet herbicide-spraying system for weed control in a lettuce field. The proposed technique involves weed/crop mapping and decision making. Experimental results show that the crop detection accuracy was 99.75%, and 98.11% of sprayable weeds were detected. The proposed technique is highly accurate, reliable and more robust than other sensor-based techniques presented in the literature. Image 1 • A real-time in-row weed-crop detection and classification algorithm was proposed. • A novel crop signaling technique was introduced to distinguish crops from weeds. • It utilized computer vision to determine the spatial location of lettuce and weed. • The algorithm was designed to equipped with a micro-jet herbicide-spraying system. • The algorithm outperformed existing methods in terms classification accuracy. [ABSTRACT FROM AUTHOR]

Evaluation was not a thing when the first author was a graduate student in the late 1970s. There was an Artificial Intelligence (AI) boom then, but that boom was quickly followed by a bust and a long AI Winter. Charles Wayne restarted funding in the mid-1980s by emphasizing evaluation. No other sort of program could have been funded at the time, at least in America. His program was so successful that these days, shared tasks and leaderboards have become common place in speech and language (and Vision and Machine Learning). It is hard to remember that evaluation was a tough sell 25 years ago. That said, we may be a bit too satisfied with current state of the art. This paper will survey considerations from other fields such as reliability and validity from psychology and generalization from systems. There has been a trend for publications to report better and better numbers, but what do these numbers mean? Sometimes the numbers are too good to be true, and sometimes the truth is better than the numbers. It is one thing for an evaluation to fail to find a difference between man and machine, and quite another thing to pass the Turing Test. As Feynman said, "the first principle is that you must not fool yourself–and you are the easiest person to fool." [ABSTRACT FROM AUTHOR]

• 3D stereo machine vision and deep learning enable pine tree architecture analysis. • Tree stem diameter, branch angle, and branch diameter are automatically estimated. • The system can assist loblolly pine breeders in selection for superior tree form. Loblolly pine is one of the most planted forest tree species in the Southern United States for sawtimber production. The sawtimber yield potential of a pine tree is significantly impacted by its stem and branch architecture, which is of important focus in tree improvement programs. However, phenotyping these traits in the upper crown of pine trees is currently based on subjective visual assessments. This study investigated the feasibility of quantifying stem diameter, branch angle, and branch diameter of six-year-old loblolly pine trees in a progeny test using stereo 3D imaging, deep learning-based instance segmentation, and image and point cloud processing techniques. Instance segmentation of branches and stems was performed as well as principal component analysis (PCA) in 2D images, followed by 3D reconstruction of the segmented organs. The resulting 3D point clouds were further processed using random sample consensus (RANSAC) and statistical outlier removal to extract stem diameter, branch angle, and branch diameter. When compared to the manual ground-truth measurements, the three system-derived parameters achieved RMSEs of 0.055 m, 5.0˚, and 5.6 mm, respectively. In addition, Bland-Altman analyses showed that the stem diameter and branch angle estimations were found with limits of agreement of ± 0.098 m and ± 9.8˚, respectively, with nonsignificant biases. On the other hand, branch diameter estimation showed −12.1 mm and 9.3 mm for lower and upper limits of agreement with a bias. The proposed system demonstrates promising potential as a high-throughput low-cost precision phenotyping tool for the characterization of loblolly pine tree architecture under field conditions, facilitating the selection of superior genotypes with improved sawtimber properties. [ABSTRACT FROM AUTHOR]

The article focuses on a study which examines large-scale observational evidence of cross-cultural differences in facial behavior using advanced in computer science. Topics discussed include the automated facial coding used, the reliability of the facial action unit detection algorithms across different testing subsets, multilevel regression estimates predicting each facial action's base rate percentage.

Some of the major restaurants and grocery chains in the United States have pledged to buy cage-free (CF) eggs only by 2025 or 2030. While CF house allows hens to perform more natural behaviors (e.g., dust bathing, perching, and foraging on the litter floor), a particular challenge is floor eggs (i.e., mislaid eggs on litter floor). Floor eggs have high chances of contamination. The manual collection of eggs is laborious and time-consuming. Therefore, precision poultry farming technology is necessary to detect floor eggs. In this study, 3 new deep learning models, that is, YOLOv5s-egg, YOLOv5x-egg, and YOLOv7-egg networks, were developed, trained, and compared in tracking floor eggs in 4 research cage-free laying hen facilities. Models were verified to detect eggs by using images collected in 2 different commercial houses. Results indicate that the YOLOv5s-egg model detected floor eggs with a precision of 87.9%, recall of 86.8%, and mean average precision (mAP) of 90.9%; the YOLOv5x-egg model detected the floor eggs with a precision of 90%, recall of 87.9%, and mAP of 92.1%; and the YOLOv7-egg model detected the eggs with a precision of 89.5%, recall of 85.4%, and mAP of 88%. All models performed with over 85% detection precision; however, model performance is affected by the stocking density, varying light intensity, and images occluded by equipment like drinking lines, perches, and feeders. The YOLOv5x-egg model detected floor eggs with higher accuracy, precision, mAP, and recall than YOLOv5s-egg and YOLOv7-egg. This study provides a reference for cage-free producers that floor eggs can be monitored automatically. Future studies are guaranteed to test the system in commercial houses. [ABSTRACT FROM AUTHOR]

• A 12-class image dataset consisting of 9370 bounding boxes was created for common weeds in cotton. • An extensive benchmark of 18 YOLO object detection models was established for weed detection. • The effect of data augmentation on weed detection was assessed. • The dataset and software programs for model benchmarking are publicly accessible. Weeds are among the major threats to cotton production. Overreliance on herbicides for weed control has accelerated the evolution of herbicide-resistance in weeds and caused increasing concerns about environments, food safety and human health. Machine vision systems for automated/robotic weeding have received growing interest towards the realization of integrated, sustainable weed management. However, in the presence of unstructured field environments and significant biological variability of weeds, it remains a serious challenge to develop reliable weed identification and detection systems. A promising solution to address this challenge are the development of arge-scale, annotated image datasets of weeds specific to cropping systems and data-driven AI (artificial intelligence) models for weed detection. Among various deep learning architectures, a diversity of YOLO (You Only Look Once) detectors is well-suited for real-time application and has enjoyed great popularity for generic object detection. This study presents a new dataset (CottoWeedDet12) of weeds important to cotton production in the southern United States (U.S.); it consists of 5648 images of 12 weed classes with a total of 9370 bounding box annotations, collected under natural light conditions and at varied weed growth stages in cotton fields. A novel, comprehensive benchmark of 25 state-of-the-art YOLO object detectors of seven versions including YOLOv3, YOLOv4, Scaled-YOLOv4, YOLOR and YOLOv5, YOLOv6 and YOLOv7, has been established for weed detection on the dataset. Evaluated through the Monte-Caro cross validation with 5 replications, the detection accuracy in terms of mAP@0.5 ranged from 88.14 % by YOLOv3-tiny to 95.22 % by YOLOv4, and the accuracy in terms of mAP@[0.5:0.95] ranged from 68.18 % by YOLOv3-tiny to 89.72 % by Scaled-YOLOv4. All the YOLO models especially YOLOv5n and YOLOv5s have shown great potential for real-time weed detection, and data augmentation could increase weed detection accuracy. Both the weed detection dataset 2 2 https://doi.org/10.5281/zenodo.7535814 and software program codes for model benchmarking in this study are publicly available 3 3 https://github.com/DongChen06/DCW , which will be to be valuable resources for promoting future research on big data and AI-empowered weed detection and control for cotton and potentially other crops. [ABSTRACT FROM AUTHOR]

Brazil is the second-largest importer, after the United States, of extra-virgin olive oils with an internal production that just reaches 0.25 % of the volume imported and consumed. Information about diagnostic chemical signatures characterizing Brazilian oils is lacking, especially as related to geographical markers and key-aroma compounds distribution. This study fills this gap; by comprehensive two-dimensional gas chromatography (GC×GC) combined with headspace solid-phase microextraction (HS-SPME) and parallel detection (MS/FID), diagnostic quali-quantitative signatures of volatiles from Brazilian oils are obtained. Comprehensively mapping of untargeted and targeted (UT) features through fingerprinting algorithms provides effective classification models, benchmarking Brazilian oils (n = 28) vs. a large selection of Italian high-quality extra-virgin olive oils (n = 111). Identitation of Brazilian oils is achieved by combining six different feature selection strategies, the resulting pattern of predictors includes: 1-methoxy-2-propanol, ethyl acetate, (E) - 4,8-dimethyl-1,3,7-nonatriene, propanoic acid, limonene, (E) - 2-hexenal, hexanoic acid, pentanal, butanoic acid, hexanal, α-pinene, (E)-β-ocimene, and propanal. Compositional differences also are evidenced in composite-class images resembling 2D patterns characteristic of the two production areas. Quantitative fingerprinting, resembling an artificial intelligence smelling machine, characterizes the unique aroma with 1-penten-3-one and (Z)− 3-hexenal, dominating with pungent and green notes, followed by (E)− 2-hexenal (bitter-almond), (E,E)− 2,4-heptadienal (fatty), eucalyptol (herbal), benzaldehyde (almond), hexanal (green), and (E)− 2-hexen-1-ol (green grass). • Pattern recognition and supervised chemometrics enable samples identitation. • UT fingerprinting and computer vision give access to high level information. • Key-aroma compounds and origin markers discriminate Italian vs. Brazilian olive oils. • Extended quantification by FID response factors is accurate and informative. • Advanced analytical tools for sensory objectification and samples discrimination. [ABSTRACT FROM AUTHOR]

We present a pattern-based regionalization of the conterminous US – a partitioning of the country into a number of mutually exclusive and exhaustive regions that maximizes the intra-region stationarity of land cover patterns and inter-region disparity between those patterns. The result is a discretization of the land surface into a number of landscape pattern types (LPTs) – spatial units each containing a unique quasi-stationary pattern of land cover classes. To achieve this goal, we use a recently developed method which utilizes machine vision techniques. First, the entire National Land Cover Dataset (NLCD) is partitioned into a grid of square-size blocks of cells, called motifels. The size of a motifel defines the spatial scale of a local landscape. The land cover classes of cells within a motifel form a local landscape pattern which is mathematically represented by a histogram of co-occurrence features. Using the Jensen–Shannon divergence as a dissimilarity function between patterns we group the motifels into several LPTs. The grouping procedure consists of two phases. First, the grid of motifels is partitioned spatially using a region-growing segmentation algorithm. Then, the resulting segments of this grid, each represented by its medoid, are clustered using a hierarchical algorithm with Ward’s linkage. The broad-extent maps of progressively more generalized LPTs resulting from this procedure are shown and discussed. Our delineated LPTs agree well with the perceptual patterns seen in the NLCD map. [ABSTRACT FROM PUBLISHER]

Despite growing concerns over the health of global invertebrate diversity, terrestrial invertebrate monitoring efforts remain poorly geographically distributed. Machine-assisted classification has been proposed as a potential solution to quickly gather large amounts of data; however, previous studies have often used unrealistic or idealized datasets to train and test their models. In this study, we describe a practical methodology for including machine learning in ecological data acquisition pipelines. Here we train and test machine learning algorithms to classify over 72,000 terrestrial invertebrate specimens from morphometric data and contextual metadata. All vouchered specimens were collected in pitfall traps by the National Ecological Observatory Network (NEON) at 45 locations across the United States from 2016 to 2019. Specimens were photographed, and two separate machine learning paradigms were used to classify them. In the first, we used a convolutional neural network (ResNet-50), and in the second, we extracted morphometric data as feature vectors using ImageJ and used traditional machine learning methods to classify specimens. Issues stemming from inconsistent taxonomic label specificity were resolved by making classifications at the lowest identified taxonomic level (LITL). Taxa with too few specimens to be included in the training dataset were classified by the model using zero-shot classification. When classifying specimens that were known and seen by our models, we reached a maximum accuracy of 72.7% using eXtreme Gradient Boosting (XGBoost) at the LITL. This nearly matched the maximum accuracy achieved by the CNN of 72.8% at the LITL. Models that were trained without contextual metadata underperformed models with contextual metadata. We also classified invertebrate taxa that were unknown to the model using zero-shot classification, reaching a maximum accuracy of 65.5% when using the ResNet-50, compared to 39.4% when using XGBoost. The general methodology outlined here represents a realistic application of machine learning as a tool for ecological studies. We found that more advanced and complex machine learning methods such as convolutional neural networks are not necessarily more accurate than traditional machine learning methods. Hierarchical and LITL classifications allow for flexible taxonomic specificity at the input and output layers. These methods also help address the 'long tail' problem of underrepresented taxa missed by machine learning models. Finally, we encourage researchers to consider more than just morphometric data when training their models, as we have shown that the inclusion of contextual metadata can provide significant improvements to accuracy. • >72,000 terrestrial invertebrate specimens spanning 3 phyla and continental scale. • We compared performance of 'traditional' machine learning and deep learning models. • XGBoost achieved very similar results to ResNet-50 CNN despite less data. • Contextual metadata improves accuracy by ∼10%. • Zero-shot classification allows for classification of unknown taxa [ABSTRACT FROM AUTHOR]

Background/Purpose Early detection of malignant melanoma is an important public health challenge. In the USA, dermatologists are seeing more melanomas at an early stage, before classic melanoma features have become apparent. Pink color is a feature of these early melanomas. If rapid and accurate automatic detection of pink color in these melanomas could be accomplished, there could be significant public health benefits. Methods Detection of three shades of pink (light pink, dark pink, and orange pink) was accomplished using color analysis techniques in five color planes (red, green, blue, hue, and saturation). Color shade analysis was performed using a logistic regression model trained with an image set of 60 dermoscopic images of melanoma that contained pink areas. Detected pink shade areas were further analyzed with regard to the location within the lesion, average color parameters over the detected areas, and histogram texture features. Results Logistic regression analysis of a separate set of 128 melanomas and 128 benign images resulted in up to 87.9% accuracy in discriminating melanoma from benign lesions measured using area under the receiver operating characteristic curve. The accuracy in this model decreased when parameters for individual shades, texture, or shade location within the lesion were omitted. Conclusion Texture, color, and lesion location analysis applied to multiple shades of pink can assist in melanoma detection. When any of these three details: color location, shade analysis, or texture analysis were omitted from the model, accuracy in separating melanoma from benign lesions was lowered. Separation of colors into shades and further details that enhance the characterization of these color shades are needed for optimal discrimination of melanoma from benign lesions. [ABSTRACT FROM AUTHOR]

Objective: Federated learning (FL) allows multiple distributed data holders to collaboratively learn a shared model without data sharing. However, individual health system data are heterogeneous. "Personalized" FL variations have been developed to counter data heterogeneity, but few have been evaluated using real-world healthcare data. The purpose of this study is to investigate the performance of a single-site versus a 3-client federated model using a previously described Coronavirus Disease 19 (COVID-19) diagnostic model. Additionally, to investigate the effect of system heterogeneity, we evaluate the performance of 4 FL variations., Materials and Methods: We leverage a FL healthcare collaborative including data from 5 international healthcare systems (US and Europe) encompassing 42 hospitals. We implemented a COVID-19 computer vision diagnosis system using the Federated Averaging (FedAvg) algorithm implemented on Clara Train SDK 4.0. To study the effect of data heterogeneity, training data was pooled from 3 systems locally and federation was simulated. We compared a centralized/pooled model, versus FedAvg, and 3 personalized FL variations (FedProx, FedBN, and FedAMP)., Results: We observed comparable model performance with respect to internal validation (local model: AUROC 0.94 vs FedAvg: 0.95, P = .5) and improved model generalizability with the FedAvg model (P < .05). When investigating the effects of model heterogeneity, we observed poor performance with FedAvg on internal validation as compared to personalized FL algorithms. FedAvg did have improved generalizability compared to personalized FL algorithms. On average, FedBN had the best rank performance on internal and external validation., Conclusion: FedAvg can significantly improve the generalization of the model compared to other personalization FL algorithms; however, at the cost of poor internal validity. Personalized FL may offer an opportunity to develop both internal and externally validated algorithms., (© The Author(s) 2022. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For permissions, please email: journals.permissions@oup.com.)

In recent decades, there have been considerable technological developments in the agriculture sector to automate manual processes for many factors, including increased production demand and in response to labor shortages/costs. We conducted a review of the literature to summarize the key advances from installing emerging technology and studies on robotics and automation to improve agricultural practices. The main objective of this review was to survey the scientific literature to identify the uses of these new technologies in agricultural practices focusing on new or reduced occupational safety risks affecting agriculture workers. We screened 3248 articles with the following criteria: (1) relevance of the title and abstract with occupational safety and health; (2) agriculture technologies/applications that were available in the United States; (3) written in English; and (4) published 2015-2020. We found 624 articles on crops and harvesting and 80 articles on livestock farming related to robotics and automated systems. Within livestock farming, most (78%) articles identified were related to dairy farms, and 56% of the articles indicated these farms were using robotics routinely. However, our review revealed gaps in how the technology has been evaluated to show the benefits or potential hazards to the safety and well-being of livestock owners/operators and workers.

• Trespassing is a safety concern for highway users and railroads at highway-rail grade crossings and along rights-of-way. • A novel vision-based trespassing detection technology was developed to automatically detect trespassing events and collect basic characteristics. • This practice-ready tool was validated and applied with over 1,600 h of video data from one grade crossing. • Data-driven practical trespassing safety risk mitigation solutions were proposed from engineering, enforcement, and education perspectives. The railroad industry plays a principal role in the transportation infrastructure and economic prosperity of the United States, and safety is of the utmost importance. Trespassing is the leading cause of rail-related fatalities and there has been little progress in reducing the trespassing frequency and deaths for the past ten years in the United States. Although the widespread deployment of surveillance cameras and vast amounts of video data in the railroad industry make witnessing these events achievable, it requires enormous labor-hours to monitor real-time videos or archival video data. To address this challenge and leverage this big data, this study develops a robust Artificial Intelligence (AI)-aided framework for the automatic detection of trespassing events. This deep learning-based tool automatically detects trespassing events, differentiates types of violators, generates video clips, and documents basic information of the trespassing events into one dataset. This study aims to provide the railroad industry with state-of-the-art AI tools to harness the untapped potential of video surveillance infrastructure through the risk analysis of their data feeds in specific locations. In the case study, the AI has analyzed over 1,600 h of archival video footage and detected around 3,000 trespassing events from one grade crossing in New Jersey. The data generated from these big video data will potentially help understand human factors in railroad safety research and contribute to specific trespassing proactive safety risk management initiatives and improve the safety of the train crew, rail passengers, and road users through engineering, education, and enforcement solutions to trespassing. [ABSTRACT FROM AUTHOR]

Built environment neighborhood characteristics are difficult to measure and assess on a large scale. Consequently, there is a lack of sufficient data that can help us investigate neighborhood characteristics as structural determinants of health on a national level. The objective of this study is to utilize publicly available Google Street View images as a data source for characterizing built environments and to examine the influence of built environments on chronic diseases and health behaviors in the United States. Data were collected by processing 164 million Google Street View images from November 2019 across the United States. Convolutional Neural Networks, a class of multi-layer deep neural networks, were used to extract features of the built environment. Validation analyses found accuracies of 82% or higher across neighborhood characteristics. In regression analyses controlling for census tract sociodemographics, we find that single-lane roads (an indicator of lower urban development) were linked with chronic conditions and worse mental health. Walkability and urbanicity indicators such as crosswalks, sidewalks, and two or more cars were associated with better health, including reduction in depression, obesity, high blood pressure, and high cholesterol. Street signs and streetlights were also found to be associated with decreased chronic conditions. Chain link fence (physical disorder indicator) was generally associated with poorer mental health. Living in neighborhoods with a built environment that supports social interaction and physical activity can lead to positive health outcomes. Computer vision models using manually annotated Google Street View images as a training dataset were able to accurately identify neighborhood built environment characteristics. These methods increases the feasibility, scale, and efficiency of neighborhood studies on health.

Wide-baseline mapping technology has been applied in NASA's Mars Exploration Rover (MER) mission to improve the accuracy of mapping of far-range terrain from Rover stereo images. As a basic research topic in photogrammetry and computer vision, it is desirable to perform a comprehensive investigation on the accuracy and automation of wide- baseline mapping. This paper presents the results of a systematic accuracy analysis of wide-baseline mapping through theoretical derivation and Monte Carlo simulation. Automated bundle adjustment and 3D wide-baseline mapping techniques are developed and tested using wide-baseline images acquired by the Spirit Rover. Experimental results demonstrate the effectiveness of the proposed techniques and verify that the mapping capability of rover stereo images can be extended from tens of meters (hard-baseline) to hundreds of meters using wide-baseline mapping techniques. [ABSTRACT FROM AUTHOR]

In the year 1999 the world population reached 6 billion, doubling the previous census estimate of 1960. Recently, the United States Census Bureau issued a revised forecast for world population showing a projected growth to 9.4 billion by 2050 (US Census Bureau, ). Different research disci- plines have studied the crowd phenomenon and its dynamics from a social, psychological and computational standpoint respectively. This paper presents a survey on crowd analysis methods employed in computer vision research and discusses perspectives from other research disciplines and how they can contribute to the computer vision approach. [ABSTRACT FROM AUTHOR]

The article provides information on the importance of being contextual in terms of an unmanned vehicle control performance in the U.S. Vehicles need to be autonomous with effective context interpretation, that is having the continuous understanding and monitoring of the external environment in sensory-based vehicle control systems. The term context is defined to understand deeply on context dependency in practical systems. The advantages of applying context monitoring to unmanned vehicle operation is discussed as well as the context-related features of visual system that affect the performance of computer vision algorithms and the adaptation of the sensing system.

The article provides information about Vision and Intelligent Systems Lab (VisLab) and the evolution of unmanned ground vehicles in the U.S. VisLab developed several research prototypes based on different technologies in collaboration with automotive partners and research centers that focuses on the protection of road users, preventive safety, enhanced vision, and military applications. Information about the pioneering projects of VisLab, the ARGO and the Surface Antartic Robot, are also offered. All the projects developed by VisLab has a common need for a real-time software system that can acquire data from different vehicle sensors and perform playback in the laboratory for algorithm development.

This article is my personal account on the work at Stanford on Stanley, the winning robot in the DARPA Grand Challenge. Between July 2004 and October 2005, my then-postdoc Michael Montemerlo and I led a team of students, engineers, and professionals with the single vision of claiming one of the most prestigious trophies in the field of robotics: the DARPA Grand Challenge (DARPA 2004). The Grand Challenge, organized by the U.S. government, was unprecedented in the nation's history. It was the first time that the U.S. Congress had appropriated a cash price for advancing technological innovation. My team won this prize, competing with some 194 other teams. Stanley was the fastest of five robotic vehicles that, on October 8, 2005, successfully navigated a 131.6-mile-long course through California's Mojave Desert. This essay is not about the technology behind our success; for that I refer the interested reader to recent articles on the technical aspects of Stanley (Dahlkamp et al. 2006; Montemerlo et al. 2006; Stavens and Thrun 2006; Thrun, Montemerlo, and Aron 2006; Thrun et al. 2006). Instead, this is my personal story of leading the Stanford Racing Team. It is the story of a team of people who built an autonomous robot in record time. It is also a success story for the field of artificial intelligence, as Stanley used some state of the art AI methods in areas such as probabilistic inference, machine learning, and computer vision. Of course, it is also the story of a step towards a technology that, one day, might fundamentally change our lives. [ABSTRACT FROM AUTHOR]

Focuses on the computer vision research for undergraduates in the United States. Ability of students to prepare and present papers in conferences; Accomplishments of students in journal publications; Continuance of students on graduate studies at various institutions.

Objective: To determine whether a computer simulation of human vision could detect and discriminate cervical cancer precursors and cancer from normal epithelium.Methods: The Georgia Tech Vision (GTV) system was trained to recognize normal and abnormal cervical features. GTV then assessed a new series of images to determine whether it could detect and diagnose cervical neoplasia without any a priori information about the images.Results: After training on one set and testing on another set of images, GTV demonstrated 100% sensitivity and 98% specificity to detect cervical intraepithelial neoplasia 3 (CIN3). GTV also had 88% sensitivity and 93% specificity to detect cervical cancer following training on one set and testing on another set of digitized cervical colposcopic images.Conclusion: This study demonstrated that GTV can detect CIN3 and cancer from digitized cervical colposcopic images. Furthermore, GTV could discriminate cervical cancer precursors and cancer from normal cervical epithelium, including immature metaplasia. [ABSTRACT FROM AUTHOR]

Objective: To investigate whether a computer vision system is comparable with humans in detecting early retinal lesions of diabetic retinopathy using color fundus photographs. Methods: A computer system has been developed using image processing and pattern recognition techniques to detect early lesions of diabetic retinopathy (hemorrhages and microaneurysms, hard exudates, and cotton-wool spots). Color fundus photographs obtained from American Indians in Oklahoma were used in developing and testing the system. A set of 369 color fundus slides were used to train the computer system using 3 diagnostic categories: lesions present, questionable, or absent (Y/Q/N). A different set of 428 slides were used to test and evaluate the system, and its diagnostic results were compared with those of 2 human experts—the grader at the University of Wisconsin Fundus Photograph Reading Center (Madison) and a general ophthalmologist. The experiments included comparisons using 3 (Y/Q/N) and 2 diagnostic categories (Y/N) (questionable cases excluded in the latter). Results: In the training phase, the agreement rates, sensitivity, and specificity in detecting the 3 lesions between the retinal specialist and the computer system were all above 90%. The κ statistics were high (0.75-0.97), indicating excellent agreement between the specialist and the computer system. In the testing phase, the results obtained between the computer system and human experts were consistent with those of the training phase, and they were comparable with those between the human experts. Conclusions: The performance of the computer vision system in diagnosing early retinal lesions was comparable with that of human experts. Therefore, this mobile, electronically easily accessible, and noninvasive computer system, could become a mass screening tool and a clinical aid in diagnosing early lesions of diabetic retinopathy. [ABSTRACT FROM AUTHOR]

Most of the existing methods focus mainly on the extraction of shape-based, rotation-based, and motion-based features, usually neglecting the relationship between hands and body parts, which can provide significant information to address the problem of similar sign words based on the backhand approach. Therefore, this paper proposes four feature-based models. The spatial-temporal body parts and hand relationship patterns are the main feature. The second model consists of the spatial-temporal finger joint angle patterns. The third model consists of the spatial-temporal 3D hand motion trajectory patterns. The fourth model consists of the spatial-temporal double-hand relationship patterns. Then, a two-layer bidirectional long short-term memory method is used to deal with time-independent data as a classifier. The performance of the method was evaluated and compared with the existing works using 26 ASL letters, with an accuracy and F1-score of 97.34% and 97.36%, respectively. The method was further evaluated using 40 double-hand ASL words and achieved an accuracy and F1-score of 98.52% and 98.54%, respectively. The results demonstrated that the proposed method outperformed the existing works under consideration. However, in the analysis of 72 new ASL words, including single- and double-hand words from 10 participants, the accuracy and F1-score were approximately 96.99% and 97.00%, respectively.

This paper explains the use of a convolutional neural network (CNN) to segment solar panels in a satellite image containing solar arrays, and extract associated metadata from the arrays. A novel unsupervised technique is introduced to estimate the azimuth of each individual solar panel from the predicted mask of the convolutional neural network. This pipeline was developed with the aim of extracting necessary metadata for a solar installation, using only a set of latitude–longitude coordinates. Azimuth prediction results for 669 individual solar installations associated with 387 sites located across the United States are provided. A mean average error and median average error of 21.65 degrees and 1.0 degrees were obtained, respectively, when predicting the azimuth of the solar fleet data set, with about 80% of the results within an error of zero degrees of the ground truth azimuth value and about 85% within an error of 25 degrees. The predicted azimuth was then used to estimate the energy conversion of the solar arrays. Results show a 90.9 and 90.6 R-squared value for estimating alternating current (AC) and direct current (DC) energy, respectively, and a mean absolute percentage error (MAPE) of 1.70% in estimating the alternating current (AC) energy using the fully automated algorithm. [Display omitted] • A novel unsupervised technique is used to estimate the azimuth of solar arrays. • Segmentation of solar panels in the satellite image, using transfer learning and a CNN. • The proposed approach is can be used with high resolution satellite or aerial imagery. [ABSTRACT FROM AUTHOR]

Focuses on the development of a machine vision technology at Virginia Polytechnic Institute & State University in Blacksburg, Virginia for the forest products industry. Color imaging system; Laser-based ranging system; X-ray scanning system; Materials handling system; Image processing system; Control computer; Machine vision software.

Technological advances in areas such as machine vision systems are accelerating the penetration of industrial robots into the U.S. workplace. Recent forecasts suggest that there may be as many as 315,000 robots in use in American industry by the close of this century. Affected managers need to design a robotics planning and implementation process compatible with their organizational climate and strategic plan. Based on a series of preliminary discussions with trade specialists, consultants, and vendors, 50 industrial firms were contacted to determine if they had implemented robotics technology. Of the 23 that had robotics on-line, nine agreed to participate in the study. All participants were personally interviewed using an open-ended questionnaire. The respondents were management or engineering personnel directly involved in the implementation process within their firm. After obtaining a description of the affected manufacturing process, the interviews focused on the factors influencing the adoption decision. Key data collected included the characteristics of the planning and adoption cycle, as well as the favorable impacts resulting from the implementation. Several important personnel-related issues, such as change in operator skill level, were discussed at length. The similarities of the respondents' planning and adoption cycles included the following: 1. Improving the firm's competitive position was consistently cited as the principal reason for the implementation. 2. Top management both initiated and supported the introduction in all cases. 3. The adoption cycle focused on a gradual implementation. 4. A robotic form consisting of an arm attached to a stationary base was the only type implemented. 5. The workers' response was judged to be positive, although several different patterns emerged. 6. All made extensive use of vendor training. The two motivating factors for the implementation were a desire to either update process technology or expand production capacity. All of the replaced operators were cross-trained in other production areas or utilized as process monitors. Quality improvement was mentioned as a benefit by seven respondents; the remaining two use robots exclusively for loading/unloading. A diverse set of other benefits was described, while the financial payback averaged one and one-half years for the five reporting firms. Selected technical implementation factors were also collected and analyzed. For example, eight respondents retrofitted the robotics technology, while seven required either upgrading or downgrading the robots' capabilities to better match the users' needs. The two respondents that did not support an inhouse awareness program did not have any human interfaces with the robots. The only respondent who did not maintain an ongoing evaluation of the new technology was the lone participant who decided against any future implementations. The analysis of the data collected lead to a series of management guidelines for planning a successful robotics implementation. These guidelines include: the need for an automation strategic plan, top management interest and support, a gradual implementation process, monitoring affected personnel, and an information system for planning and control. [ABSTRACT FROM AUTHOR]

• Sows in wider crates sat less and laid down more. • Piglets with an additional heat lamp spent more time in the heated area. • Piglet location in farrowing stalls was correlated with sow lying orientation. • Piglet mortality was correlated with lactation day, sow lying, standing and feeding. • Piglet location was not correlated with piglet pre-weaning mortality. Farrowing stalls are used in the United States swine industry to reduce pre-weaning piglet mortality, enable efficient individual animal management, and decrease facility construction and operating costs. The quantity and quality of space provided for sows and piglets in farrowing stalls are important economic and welfare considerations. To further explore the impacts of farrowing stall space allocation, a large-scale field study was conducted to compare sow and piglet behavior when housed in three farrowing stall layouts (TSL – traditional stall layout, ECSL – expanded creep area stall layout, ESCSL – expanded sow and creep area stall layout) with either one or two heat lamps (1HL and 2HL, respectively). A computer vision system classified posture budgets and behaviors of 322 sows and piglet location for 324 litters. Linear mixed models were developed to compare behavior and piglet pre-weaning mortality metrics between experimental treatments. Results show sows in ESCSL spent more time lying compared to sows in ECSL (p = 0.028) and less time sitting compared to sows in TSL and ECSL (p < 0.01). Sows with the 2HL treatment had an increase in percentage lying (p = 0.017) and a decrease in percentage standing (p = 0.045) compared to sows with the 1HL treatment. Number of piglets, parity, and batch also influenced sow postural behavior (p < 0.05). Sow lying orientation was not impacted by HL treatment. Sow postures and behaviors were influenced by day of lactation (p < 0.001). Piglets with 2HL treatment spent more time in the heated region and less time in the creep and sow regions for all stall layouts on all days of lactation observed (p < 0.001). In the ESCSL, piglets had a greater percentage of time in the sow region compared to ECSL piglets (p < 0.004). Piglets did not spend equal percentages of time between the two creep or two HL regions (p < 0.001), and piglet location was correlated with sow lying orientation for most of the creep regions analyzed (p < 0.01). Increases in piglet pre-weaning mortality were correlated with increases in sow lying (p = 0.027) and decreases in standing (p = 0.025) and feeding (p < 0.001). However, correlations with sow posture were likely due to the impacts of day of lactation (p < 0.001). No correlations were found between piglet location and pre-weaning mortality (p> 0.05). Results can guide producers to consider wider sow areas in farrowing stalls to better meet sow behavioral needs and to include larger heated areas to meet piglet behavioral needs during lactation. [ABSTRACT FROM AUTHOR]