Visual transformers have driven major progress in remote sensing image analysis, particularly in object detection and segmentation. Recent vision-language and multimodal models further extend these capabilities by incorporating auxiliary information, including captions, question and answer pairs, and metadata, which broadens applications beyond conventional computer vision tasks. However, these models are typically optimized for semantic alignment between visual and textual content rather than geospatial understanding, and therefore are not suited for representing or reasoning with structured geospatial layers. In this study, we propose a novel model that enhances remote sensing imagery processing with guidance from auxiliary geospatial information. Our approach introduces a geospatial embedding mechanism that transforms diverse geospatial data into embedding patches that are spatially aligned with image patches. To facilitate cross-modal interaction, we design a guided attention module that dynamically integrates multimodal information by computing attention weights based on correlations with auxiliary data, thereby directing the model toward the most relevant regions. In addition, the module assigns distinct roles to individual attention heads, allowing the model to capture complementary aspects of the guidance information and improving the interpretability of its predictions. Experimental results demonstrate that the proposed framework outperforms existing pretrained geospatial foundation models in predicting disease prevalence, highlighting its effectiveness in multimodal geospatial understanding.

This paper presents the results of the fourth edition of the Monocular Depth Estimation Challenge (MDEC), which focuses on zero-shot generalization to the SYNS-Patches benchmark, a dataset featuring challenging environments in both natural and indoor settings. In this edition, we revised the evaluation protocol to use least-squares alignment with two degrees of freedom to support disparity and affine-invariant predictions. We also revised the baselines and included popular off-the-shelf methods: Depth Anything v2 and Marigold. The challenge received a total of 24 submissions that outperformed the baselines on the test set; 10 of these included a report describing their approach, with most leading methods relying on affine-invariant predictions. The challenge winners improved the 3D F-Score over the previous edition's best result, raising it from 22.58% to 23.05%.

Local Binary Pattern (LBP) and its variants have considerable success in a wide range of computer vision and pattern recognition applications, especially in tasks related to texture classification. However, the LBP method is sensitive to noise, scale variations and unable to capture macro-structure information. We propose a novel texture classification descriptor called Scale Adaptive Robust LBP (SARLBP) that enhances macro-level descriptive information by incorporating significantly larger scales, and a novel encoding scheme, which is designed to overcome the limitations of traditional LBP schemes. SARLBP method dynamically determines a single optimal scale for each radial direction from multiple scales based on the local area’s characteristics. Subsequently, this descriptor extracts four distinct patterns derived from regional image medians of center pixel, radially-optimized neighbor pixels, optimized fixed scale-based pixels, and radial-difference-based pixels. This method adeptly captures texture information at both micro and macro scales by employing scale adaptation based on the distinctive attributes of the local region. As a result, it provides a comprehensive and robust representation of the texture images. Extensive experimentation was conducted on four publicly available texture databases (ALOT, CUReT, UMD, and Kylberg), considering both the presence and absence of two distinct types of interference (Gaussian noise and Salt-and-Pepper noise). The results reveal that our SARLBP method achieves significantly better performance than other state-of-the-art LPB variants with a fixed smaller feature dimension.

OBJECTIVES:Relative fundamental frequency (RFF) estimates laryngeal tension during speech, providing insights into vocal effort. Current methods to derive RFF from continuous speech require manual processing, hindering large-scale studies with ecologically valid speech productions. This research aimed to develop and evaluate three fully automated pipelines for RFF analysis from continuous speech, addressing this limitation. METHODS:Three pipelines were compared: two modifications of an existing semiautomated approach [automated relative fundamental frequency (aRFF)-AP] and one novel pipeline replicating manual analysis. The pipelines were tested on speech samples containing vowel-consonant-vowel (VCV) utterances from 82 female participants with and without vocal fatigue complaints in the absence of phonotraumatic vocal fold changes. The pipelines automatically segmented VCVs and measured RFF. Manual measurements of a subset provided reliability and validity benchmarks. RESULTS:All pipelines demonstrated good reliability (r ≥ 0.84) and validity when compared with manual analysis. They required minimal manual correction (<4%) for fricative identification. Notably, the novel aRFF-B pipeline rejected the fewest samples (10%-25%) while maintaining reliability and was able to leverage parallel computing. CONCLUSIONS:Three automated pipelines, especially aRFF-B, enabled time-efficient RFF analysis of large continuous speech data sets without manual intervention. This advancement can facilitate large-scale studies using RFF applied to continuous speech, potentially expanding its application in voice research and clinical practice.

Data-Driven Inverse Kinematics (DDIK) solvers emerged as promising Inverse Kinematics (IK) methods for reliably approximating the IK of robotic manipulators. However, these solvers remain heavily robot-dependent, where for each robot of interest, a network needs to be trained in an one-solver-one-robot framework. In this paper, we build on our previous work on Learning-By-Example (LBE) for DDIK, and introduce an one-solver-many-robots framework; where a single neural network is used to predict the IK of multiple robots - mainly with 6 and 7 Degrees of Freedom (DoF). In our LBE approach, the neural network input includes an example of joint-pose tuple (e.g. any previous joint and corresponding pose tuple in the path) along with the queried pose as the same network outputs the desired robot joint configuration. Here, we investigate five network architectures: a Plain Multilayer Perceptron (MLP), a Residual-based MLP (RMLP), a Densely Connected MLP (DMLP), and two transformers inspired by Generative Pre-trained Transformer (GPT) and tested them using 3 diverse datasets with 20 real-world robotic arms with 6 and 7 DoF. Our experimental results demonstrate that a single lightweight, LBE-based DDIK solver can reliably predict the IK for multiple and hitherto unseen robots, within each of the 6 or 7DoF family as well as across both 6 and 7DoF robot families with position errors below 1mm and orientation errors below 1deg. Additionally, we compare all proposed LBE-DDIK solvers with three established numerical IK solvers: Selectively Damped Least-Squares (SD), Singular Value Filtering (SVF), and Mixed Inverse (MX) and observe that our LBE-DDIK solvers achieve comparable accuracy, with the advantage of being a one-solver-many-robots framework.

Nuanced-concept image classification tasks often require substantial labeled data. The labeling process for such problems is time-consuming and labor-intensive. While zero-shot methods like Modeling Collaborator have shown promising results, they generally lack a versatile open source pipeline for domain-independent, multi-class fine-grained classification. We are proposing a classification pipeline consisting of weak supervision and opensource Vision Language Models (VLMs) to be employed in both binary and multi-class nuanced classification problems. Our proposed pipeline is domain-independent as it uses knowledge embedded in the pre-training of VLMs. This eliminates the need for additional fine-tuning for specific contexts, as required by methods such as AdaptCLIPZS. In our proposed pipeline, VLMs serve as weak labelers in the classification tasks, while a Weak Supervision (WS) model aggregates those labels and produce a set of pseudo labels (pseudo ground-truth) to train an end classifier. We have conducted multiple experiments to demonstrate the validity of the pipeline in both binary and multi-class classification tasks. The experimental results have shown that our proposed pipeline is capable of producing superior results in both binary and multi-class problems compared to the state-of-the-art zero-shot classification methods.

Manual analysis and diagnosis of COVID-19 through the examination of Computed Tomography (CT) images of the lungs can be time-consuming and result in errors, especially given high volume of patients and numerous images per patient. So, we address the need for automation of this task by developing a new deep learning-based pipeline. Our motivation was sparked by the CVPR Workshop on "Domain Adaptation, Explainability and Fairness in AI for Medical Image Analysis", more specifically, the "COVID-19 Diagnosis Competition (DEF-AI-MIA COV19D)" under the same Workshop. This challenge provides an opportunity to assess our proposed pipeline for COVID-19 detection from CT scan images. The same pipeline incorporates one of the architectures in the EfficientNet "family", but with an added Spatial Attention Mechanism: EfficientNet-SAM. Also, unlike the traditional/past pipelines, which relied on a preprocessing step, our pipeline takes the raw selected input images without any such step, except for an image-selection step to simply reduce the number of CT images required for training and/or testing. Moreover, our pipeline is computationally efficient, as, for example, it does not incorporate a decoder for segmenting the lungs. It also does not combine different models nor combine RNN with a backbone, as other pipelines in the past did. Nevertheless, our pipeline outperformed all approaches presented by other teams in last year's instance of the same challenge using the validation subset. It also placed 5th in this year's competition, ranking less than 1.3% below the 1st place and close to 3.5% above the 6th place based on the macro-F1 score.

Outlier interpretation is essential in various data processing fields within the context of outlier detection. Under-standing the specific attributes that make an outlier distinct is particularly valuable in many application areas, especially where, due to class imbalance, it can greatly enhance analytic insights into unusual occurrences or trends. To address this need, we introduce a Regularized Auto-Encoder with a Genetic Algorithm (RAE-GA) approach for outlying and characterizing subspaces: i.e. a novel approach for identifying outliers and pinpointing their underlying subspaces. This method leverages autoencoders with regularized latent spaces, such as in Variational Auto Encoders (VAE) and Regularized Auto Encoders (RAE), which have proven effective in outlier detection only. So, building upon this success, the proposed RAE-GA utilizes a specially crafted, computationally efficient fitness function based on RAE. This function is integrated into the genetic algorithm to explore various potential subspaces, eliminating the need for restrictive pruning or constraints. As our experimental results indicate, RAE-GA outperforms the current state-of-the-art methods in outlier interpretation.

We introduce a high-throughput 3D scanning system designed to accurately measure cattle phenotypes. This scanner employs an array of depth sensors, i.e., time-of-flight (ToF) sensors, each controlled by dedicated embedded devices. The sensors generate high-fidelity 3D point clouds, which are automatically stitched using a point could segmentation approach through deep learning. The deep learner combines raw RGB and depth data to identify correspondences between the multiple 3D point clouds, thus creating a single and accurate mesh that reconstructs the cattle geometry on the fly. In order to evaluate the performance of our system, we implemented a two-fold validation process. Initially, we quantitatively tested the scanner for its ability to determine accurate volume and surface area measurements in a controlled environment featuring known objects. Next, we explored the impact and need for multi-device synchronization when scanning moving targets (cattle). Finally, we performed qualitative and quantitative measurements on cattle. The experimental results demonstrate that the proposed system is capable of producing high-quality meshes of untamed cattle with accurate volume and surface area measurements for livestock studies.

Classification and segmentation using ultra-fine-grained datasets can be challenging due to the small nuances between adjacent classes. This problem can be exacerbated by the fact that variations within classes can be much larger than other variations between classes. Some approaches have resorted to attention mechanisms that focus on the source or the properties of the features that cause these minor changes in samples between or within classes. In some cases, the attention mechanism can be derived from spatial, temporal, modal, or other types of features in the dataset. Sometimes, attention can be drawn from external sources such as the shape of the object, its skeleton, contour, etc. Finally, some approaches use completely independently extracted information to guide the attention mechanism in a supervised fashion (privileged information, guided-attention, etc). In this paper, we claim that in the context of ultra-fine datasets with a small number of samples, a simple attention mechanism can improve the classification results. Moreover, the same simple attention mechanism can be employed in a backbone topology for the segmentation of the same information that would otherwise be used to guide the attention mechanism in other methods. In other words, unlike the state-of-the-art model for ultra-fine-grained classification of, for example, plant leaves datasets, which uses segmentation masks to guide its attention mechanism, our proposed network can simultaneously provide a classification label and a segmentation mask. The XGBoost algorithm was applied to the attention-modulated feature map for classification, and the Optuna hyperparameter optimization framework was used to tune XGBoost. Three state-of-the-art methods were compared against ours using three benchmark datasets, and our model, XMNet, achieved the best results for the vein segmentation task. For the classification part, our network achieved comparable performance with respect to two state-of-the-art as well as various other more traditional methods.

Highlights A transfer learning strategy improved residue estimates from high-resolution RGB imagery. The best method used probabilistic estimates of expert classifiers to estimate residue cover. This research confirms the utility of RGB imagery to quantify residue cover in agricultural fields. Abstract. Plant residue on the soil surface increases the sustainability of food and fiber production in agricultural systems. Automated assessments of residue cover based on imagery have the potential to reduce labor and human bias associated with in-field measurements. We evaluate the capacity of a transfer learning strategy to improve the determination of residue level from high-resolution RGB images. The imagery for the study was collected from 88 field locations in 40 row crop fields in five Missouri counties between mid-April and early July in 2018 and 2019. At each field location, 50 contiguous 0.3 m × 0.2 m region of interest (ROI) images (ground sampling distance of 0.014 cm pixel--1) were extracted from imagery, resulting in a dataset of 4,400 ROI images; 3,000 were used for cross-validation and training (data collected in 2018) and 1,400 were used for testing (data collected in 2019). The percentage residue for each ROI image (ground truth) was determined by a bullseye grid method (n = 100). Features were extracted from ROI images using the VGGNet-16 CNN model, a pre-trained convolutional neural network model. We extracted 1,472 features per ROI using a global averaging and pooling strategy. The optimum feature set was identified using recursive feature elimination using a support vector machine (RFE-SVM). To estimate crop residue percentage using selected features, expert two-class SVMs were trained to separate adjacent levels of residue cover, where the rationale of the ensemble was to allow each of the two-class SVMs to find the hyperplanes that maximize the margin between the corresponding two consecutive classes. Based on the distance of the samples to these hyperplanes, probabilistic estimates of the data-point belonging to the class were computed. With the combined knowledge of probabilistic estimates from each expert classifier, the percentage crop residue cover of each ROI image was calculated. We tested our approach with 3-, 4-, 5-, and 8-class problems, achieving the best results with the 8-class problem with r2 = 0.93 at the ROI level, r2 = 0.97 at the field-location level, and minimal bias in residue estimates in low residue conditions. These results are superior to other reported estimates of percent residue derived from imagery. This research confirms the utility of high-resolution RGB imagery to quantify residue cover in agricultural systems. Keywords: Convolutional neural network, Soil erosion, Support vector machine, Transfer learning.

Inverse Kinematics (IK) is one of the most fundamental challenges in robotics. It refers to the process of determining the joint configurations required to achieve the desired position and orientation (pose) of a robot end-effector. Although numerous Data-Driven (DD) IK solvers have demonstrated encouraging results, they have not achieved the same accuracy when compared to other IK methods for complex robot configurations (e.g., numerical methods for higher Degrees of Freedom (DoF)). In this work, we propose a new Learning-by-Example method, and show that such a scheme considerably improves the IK learning results when compared to other DD learners. In our approach, the network input incorporates an example of joint-pose pair along with the query pose to predict the desired robot joint configuration. We show that the example joint-pose pair does not need to be too close to the query - i.e. example and query can be as far as 20 degrees apart in the joint configuration space. Furthermore, we investigate the utilization of residual and dense skip connections in Multilayer Perceptron for DDIK solvers and employ the resulting networks for two redundant robotic manipulators: a 7-DoF-7R commensurate robot and a 7-DoF-2RP4R incommensurate robot. Our experimental results show that the resulting DDIK solver can reliably predict IK solutions with accuracy better than 1mm in position and 1deg in orientation.

Performing feature selection on a small number of instances with high-dimensional datasets poses a needed challenge in preventing over-fitting. To address this issue, this paper proposes a sequential transfer-learning approach combined with a multi-objective genetic algorithm (STMO-GA) for feature selection. Firstly, for the multi-objective component of our method, we employ a Non-dominated Sorting Genetic Algorithm (NSGA-II) to generate a Pareto front. Then, features are ranked based on their number of appearances in the same Pareto front. Next, during the sequential knowledge transfer process, the ranked features are iteratively selected until a predetermined $n$ number of features remains. This feature subspace is further refined by a k-fold cross-validation operation, starting from the rank-one feature, to determine the cut-off of the $n$ features that will remain. Comparative evaluations against both GA-based as well as traditional feature selection methods demonstrate that the proposed method achieves superior classification accuracy, while retaining the smallest number or a comparable number of features.

Violence against women (VAW) is a serious violation of the rights to life, health, and physical integrity. Recent studies point out that social, economic, and demographic factors directly impact the advance of this type of violence. In view of these facts, the state has its responsibility increased when it cannot provide the public equipment necessary for management strategies that collaborate with the confrontation of violence. This project aims to develop a multicriteria decision analysis model (MCDA) to classify Pará municipalities with regard to the propensity for VAW crime, based on the mapping of assistance and protection equipment, as well as socioeconomic indicators of each municipality. The model developed and the research findings represent an important step in elaboration. In turn, this model demonstrates its ability to be a possible instrument that decision makers and implementers of public policies aimed at protecting and supporting women victims of violence in order to anticipate new occurrences.

Numerical methods for Inverse Kinematics (IK) employ iterative, linear approximations of the IK until the end-effector is brought from its initial pose to the desired final pose. These methods require the computation of the Jacobian of the Forward Kinematics (FK) and its inverse in the linear approximation of the IK. Despite all the successful implementations reported in the literature, Jacobian-based IK methods can still fail to preserve certain useful properties if an improper matrix inverse, e.g. Moore-Penrose (MP), is employed for incommensurate robotic systems. In this paper, we propose a systematic, robust and accurate numerical solution for the IK problem using the Mixed (MX) Generalized Inverse (GI) applied to any type of Jacobians (e.g., analytical, numerical or geometric) derived for any commensurate and incommensurate robot. This approach is robust to whether the system is under-determined (less than 6 DoF) or over-determined (more than 6 DoF). We investigate six robotics manipulators with various Degrees of Freedom (DoF) to demonstrate that commonly used GI's fail to guarantee the same system behaviors when the units are varied for incommensurate robotics manipulators. In addition, we evaluate the proposed methodology as a global IK solver and compare against well-known IK methods for redundant manipulators. Based on the experimental results, we conclude that the right choice of GI is crucial in preserving certain properties of the system (i.e. unit-consistency).

As machine learning solutions become increasingly more ubiquitous in medical diagnosis, researchers are becoming equally more aware of the possibility of confounded predictions being produced by these same models. This realization derives, for example, from the observation that sample-wise crossvalidation leads to highly underestimated error predictors when compared to subject-wise cross validation. However, without a reliable approach to remove spurious, confounding factors such as age, gender, or even the type/brand of equipment used, these same machine learning solutions will be fated to produce poor results despite the metrics for error estimation employed. In this research, we propose an optimization approach, using genetic algorithms, to adapt the feature vectors in order to maximize the prediction accuracy of a given classifiers, while minimizing the correlation between the features and the potential confounding factors. Our results, when applied to the diagnostic of vocal fatigue, have shown great improvement in terms of the generalization capability of the chosen SVM classifier. The system was evaluated using subject-wise (i.e. leave-one-subject-out) cross-validation, which demonstrated the effectiveness of this new confounding removal approach.

In this paper, we review and compare several velocity-level and acceleration-level Pseudo-Inverse-based Path Planning (PPP) and Pseudo-Inverse-based Repetitive Motion Planning (PRMP) schemes based on the kinematic model of robotic manipulators. We show that without unit consistency in the pseudo-inverse computation, path planning of incommensurate robotic manipulators will fail. Also, we investigated the robustness and noise tolerance of six PPP and PRMP schemes in the literature against various noise types (i.e. zero, constant, time-varying and random noises). We compared the simulated results using two redundant robotic manipulators: a 3DoF (2RP), and a 7DoF (2RP4R). These experimental results demonstrate that the improper Generalized Inverse (GI) with arbitrary selection of unit and/or in the presence of noise can lead to unexpected behavior of the robot, while producing wrong instantaneous outputs in the task space, which results in distortions and/or failures in the execution of the planned path. Finally, we propose and demonstrate the efficacy of the Mixed Inverse (MX) as the proper GI to achieve unit-consistency in path planning.

Highlights A machine learning framework estimated residue cover in RGB images taken at three resolutions from 88 locations. The best results primarily used texture features, the RFE-SVM feature selection method, and the SVM classifier. Accounting for shadows and plants plus modifying and optimizing the texture features may improve performance. An automated system developed using machine learning is a viable strategy to estimate residue cover from RGB images obtained with handheld or UAV platforms. Abstract . Maintaining plant residue on the soil surface contributes to sustainable cultivation of arable land. Applying machine learning methods to RGB images of residue could overcome the subjectivity of manual methods. The objectives of this study were to use supervised machine learning while identifying the best feature selection method, the best classifier, and the most effective image feature types for classifying residue levels in RGB imagery. Imagery was collected from 88 locations in 40 row-crop fields in five Missouri counties between early May and late June in 2018 and 2019 using a tripod-mounted camera (0.014 cm pixel-1 ground sampling distance, GSD) and an unmanned aerial vehicle (UAV, 0.05 and 0.14 GSD). At each field location, 50 contiguous 0.3 × 0.2 m region of interest (ROI) images were extracted from the imagery, resulting in a dataset of 4,400 ROI images at each GSD. Residue percentages for ground truth were estimated using a bullseye grid method (n = 100 points) based on the 0.014 GSD images. Representative color, texture, and shape features were extracted and evaluated using four feature selection methods and two classifiers. Recursive feature elimination using support vector machine (RFE-SVM) was the best feature selection method, and the SVM classifier performed best for classifying the amount of residue as a three-class problem. The best features for this application were associated with texture, with local binary pattern (LBP) features being the most prevalent for all three GSDs. Shape features were irrelevant. The three residue classes were correctly identified with 88%, 84%, and 81% 10-fold cross-validation scores for the 2018 training data and 81%, 69%, and 65% accuracy for the 2019 testing data in decreasing resolution order. Converting image-wise data (0.014 GSD) to location residue estimates using a Bayesian model showed good agreement with the location-based ground truth (r2 = 0.90). This initial assessment documents the use of RGB images to match other methods of estimating residue, with potential to replace or be used as a quality control for line-transect assessments. Keywords: Feature selection, Soil erosion, Support vector machine, Texture features, Unmanned aerial vehicle.

Supplemental Table S1. Libraries used for each step of the machine learning operations.

Our previous studies demonstrated that it is possible to perform the classification of both simulated pressed and actual vocally fatigued voice productions versus vocally healthy productions through the pattern recognition of sEMG signals obtained from subjects' anterior neck. In these studies, the commonly accepted Vocal Fatigue Index factor 1 (VFI-1) was used for the ground-truth labeling of normal versus vocally fatigued voice productions. Through recent experiments, other factors with potential effects on classification were also studied, such as sEMG signal normalization, and data imbalance-i.e., the large difference between the number of vocally healthy subjects and of those with vocal fatigue. Therefore, in this paper, we present a much improved classification method derived from an extensive study of the effects of such extrinsic factors on the classification of vocal fatigue. The study was performed on a large number of sEMG signals from 88 vocally healthy and fatigued subjects including student teachers and teachers and it led to important conclusions on how to optimize a machine learning approach for the early detection of vocal fatigue.