Uncontrolled Environment Research Articles

Plant Leaf Disease Detection Using Deep Learning: A Multi-Dataset Approach

Agricultural productivity is increasingly threatened by plant diseases, which can spread rapidly and lead to significant crop losses if not identified early. Detecting plant diseases accurately in diverse and uncontrolled environments remains challenging, as most current detection methods rely heavily on lab-captured images that may not generalise well to real-world settings. This paper aims to develop models capable of accurately identifying plant diseases across diverse conditions, overcoming the limitations of existing methods. A combined dataset was utilised, incorporating the PlantDoc dataset with web-sourced images of plants from online platforms. State-of-the-art convolutional neural network (CNN) architectures, including EfficientNet-B0, EfficientNet-B3, ResNet50, and DenseNet201, were employed and fine-tuned for plant leaf disease classification. A key contribution of this work is the application of enhanced data augmentation techniques, such as adding Gaussian noise, to improve model generalisation. The results demonstrated varied performance across the datasets. When trained and tested on the PlantDoc dataset, EfficientNet-B3 achieved an accuracy of 73.31%. In cross-dataset evaluation, where the model was trained on PlantDoc and tested on a web-sourced dataset, EfficientNet-B3 reached 76.77% accuracy. The best performance was achieved with the combination of the PlanDoc and web-sourced datasets resulting in an accuracy of 80.19% indicating very good generalisation in diverse conditions. Class-wise F1-scores consistently exceeded 90% for diseases such as apple rust leaf and grape leaf across all models, demonstrating the effectiveness of this approach for plant disease detection.

Emotion Recognition from EEG Signals Using Advanced Transformations and Deep Learning

Affective computing aims to develop systems capable of effectively interacting with people through emotion recognition. Neuroscience and psychology have established models that classify universal human emotions, providing a foundational framework for developing emotion recognition systems. Brain activity related to emotional states can be captured through electroencephalography (EEG), enabling the creation of models that classify emotions even in uncontrolled environments. In this study, we propose an emotion recognition model based on EEG signals using deep learning techniques on a proprietary database. To improve the separability of emotions, we explored various data transformation techniques, including Fourier Neural Networks and quantum rotations. The convolutional neural network model, combined with quantum rotations, achieved a 95% accuracy in emotion classification, particularly in distinguishing sad emotions. The integration of these transformations can further enhance overall emotion recognition performance.

Prosodic changes with age: a longitudinal study with three public figures in European Portuguese

Purpose The analysis of acoustic parameters contributes to the characterisation of human communication development throughout the lifetime. The present paper intends to analyse suprasegmental features of European Portuguese in longitudinal conversational speech samples of three male public figures in uncontrolled environments across different ages, approximately 30 years apart. Participants and methods Twenty prosodic features concerning intonation, intensity, rhythm, and pause measures were extracted semi-automatically from 360 speech intervals (3-4 interviews from each speaker x 30 speech intervals x 3 speakers) lasting between 3 to 6 s. Results Twelve prosodic parameters presented significant age effects at least in one speaker. Group mean comparisons revealed significant differences between the youngest (i.e. 50 years) and the oldest age groups (i.e. 80 years) in seven parameters. The results from the analysis point to a lower and less variable fo, higher fo minimum, wider fo peaks, more vocal effort and more variable global intensity, slower speech and articulation rate, and also more frequent and longer pauses in older ages. Conclusion This longitudinal study has the potential to contribute to the characterization of the normal aging process, proving to be significant in the domains of human-machine communication, speech recognition systems, applied linguistics, or the implementation of strategies in communicative contexts with older adults

The role of tourist satisfaction in mediating destination loyalty: Empirical evidence on natural dye batik in Indonesia

Tourist satisfaction shows the satisfaction that arises because visitors feel satisfied and happy after visiting a destination. This study aims to examine the role of tourist satisfaction in influencing destination image, emotional experience, and value perception towards destination loyalty of natural dye batik in Indonesia. The research uses a quantitative approach in an uncontrolled environment, with individuals as the unit of analysis. The population consisted of domestic and foreign tourist respondents aged 21 years and over who visited the natural dye batik village, Bayat, Klaten, Indonesia. Data collection was carried out using an online survey method, questionnaires were sent to selected respondents using online Google Forms. The data were collected comprehensively from 200 respondents for descriptive demographic profile analysis. The hypotheses were tested using partial least squares modelling analysis. The results of the study showed a positive influence of destination image, emotional experience, and perceived value on tourist satisfaction (Beta = .423, p < .000; Beta = .087, p < .045, Beta = .368, p < .000). Likewise, destination image, perceived value, and tourist satisfaction have a positive effect on destination loyalty (Beta = .306, p < .000; Beta = .198, p < .009, and Beta = .325, p = .000). This shows that the better the destination image and value perception, and the higher the emotional experience, the higher the tourist satisfaction, and with increasing tourist satisfaction, destination loyalty also increases. AcknowledgmentWe would like to thank the management of natural dyed batik “Kebon Indah”, Bayat Klaten (Mrs. Dalmini and staff) who permitted and facilitated this study, especially in obtaining respondents. We would like to thank our friends at the Udinus Faculty of Economics and Business who provided input to improve the results and reporting of this study and all respondents who directly or indirectly contributed to this research. Hopefully, this study will be useful for developing science and business practice.

Facial Biosignals Time–Series Dataset (FBioT): A Visual–Temporal Facial Expression Recognition (VT-FER) Approach

Visual biosignals can be used to analyze human behavioral activities and serve as a primary resource for Facial Expression Recognition (FER). FER computational systems face significant challenges, arising from both spatial and temporal effects. Spatial challenges include deformations or occlusions of facial geometry, while temporal challenges involve discontinuities in motion observation due to high variability in poses and dynamic conditions such as rotation and translation. To enhance the analytical precision and validation reliability of FER systems, several datasets have been proposed. However, most of these datasets focus primarily on spatial characteristics, rely on static images, or consist of short videos captured in highly controlled environments. These constraints significantly reduce the applicability of such systems in real-world scenarios. This paper proposes the Facial Biosignals Time–Series Dataset (FBioT), a novel dataset providing temporal descriptors and features extracted from common videos recorded in uncontrolled environments. To automate dataset construction, we propose Visual–Temporal Facial Expression Recognition (VT-FER), a method that stabilizes temporal effects using normalized measurements based on the principles of the Facial Action Coding System (FACS) and generates signature patterns of expression movements for correlation with real-world temporal events. To demonstrate feasibility, we applied the method to create a pilot version of the FBioT dataset. This pilot resulted in approximately 10,000 s of public videos captured under real-world facial motion conditions, from which we extracted 22 direct and virtual metrics representing facial muscle deformations. During this process, we preliminarily labeled and qualified 3046 temporal events representing two emotion classes. As a proof of concept, these emotion classes were used as input for training neural networks, with results summarized in this paper and available in an open-source online repository.

Predicting the Effects of Ozone on Long-Term Growth of Aspen Trees Using Response Functions Developed From Seedlings Grown in Field Chambers.

Tropospheric ozone (O3) is among the most pervasive and harmful air pollutants known to affect ecosystems. In the United States, the Environmental Protection Agency and other agencies are tasked with protecting plants and ecosystems from harmful O3 exposures. Controlled exposure experiments conducted in field open-top chambers (OTCs) with small tree seedlings have been used to estimate empirical models of tree growth in response to O3 exposure for more than 16 species. While this experimental method makes it possible to obtain detailed exposure-response data, it remains uncertain whether predictions of empirical models parameterized using those data are sufficiently accurate when applied to trees grown in uncontrolled natural environments for long periods. We used O3 exposure-response relationships developed from several OTC studies of trembling aspen (Populus tremuloides Michx.) seedlings to predict the growth of the same species in the Aspen FACE "free-air" O3 exposure experiment in Rhinelander, Wisconsin, over 11 years. We acquired individual tree growth data and hourly O3 exposure from the ambient and elevated O3 plots in the Aspen FACE experiment, computed annual exposure using the same metrics of O3 exposure as were used in the OTC seedling experiments, and generated predictions of growth in the Aspen FACE exposures. A simple empirical model parameterized using the OTC seedling data accurately predicted the percent above-ground biomass loss due to O3 exposure in the Aspen FACE trees for all 11 years. In the Aspen FACE experiment, the effect of O3 exposure was established in early years and continued to be observed in later years without worsening. Our study suggests that O3 exposure-response relationships obtained from OTC seedling studies can be used to make inferences about effects on larger trees. These results imply that researchers can use these relationships with confidence when estimating risks of O3 pollution across the United States.

PNasFH-Net: Pyramid Neural Architecture Search Forward Network for Facial Emotion Recognition in Uncontrolled and Pose Variant Environment

PNasFH-Net: Pyramid Neural Architecture Search Forward Network for Facial Emotion Recognition in Uncontrolled and Pose Variant Environment

Unraveling the Applicability of LbL Coatings for Drug Delivery in Dental Implant-Related Infection Treatment.

Peri-implantitis is an inflammatory condition caused by bacterial biofilms adhered on dental implant surfaces that cause progressive tissue destruction from the host's inflammatory response. The adverse effects of peri-implantitis progression can go beyond just losing the implant. This highlights the importance of implementing strategies to stabilize disease in the short term. Layer-by-layer (LbL) assembly is a promising avenue in the field of peri-implantitis management due to its applicability with a variety of substances, in addition to being an easy, versatile, and flexible process for multilayer formation to act directly in the affected site. In this Review, our objective is to offer comprehensive chemical and biological insights into the LbL system, clarifying its specific application as antimicrobial coatings, with concern for the physical site and purpose. Additionally, we delve deeper into the concepts of onset and progression of peri-implantitis, aiming to elucidate the precise indications for employing the LbL system as a coating for implant abutments in peri-implantitis treatment. Finally, we correlate the chemical composition of the LbL system with its functionality while also addressing the challenges posed by the uncontrolled environment of the oral cavity, which ultimately restricts its clinical applicability.

Severity Classification of Parkinson's Disease via Synthesis of Energy Skeleton Images from Videos Produced in Uncontrolled Environments.

Background/Objectives: Parkinson's Disease is a prevalent neurodegenerative disorder affecting millions worldwide, primarily marked by motor and non-motor symptoms due to the degeneration of dopamine-producing neurons. Despite the absence of a cure, current treatments focus on symptom management, often relying on pharmacotherapy and surgical interventions. Early diagnosis remains a critical challenge, particularly in underserved areas, as existing diagnostic protocols lack standardization and accessibility. This paper proposes a novel framework for the diagnosis and severity classification of PD using video data captured in uncontrolled environments. Methods: Leveraging deep learning techniques, our approach synthesizes Skeleton Energy Images (SEIs) from gait sequences and employs three advanced models-a Convolutional Neural Network (CNN), a Residual Network (ResNet), and a Vision Transformer (ViT)-to analyze these images. Our methodology allows for the accurate detection of PD and differentiation of its severity without requiring specialized equipment or professional oversight. The dataset used consists of labeled videos capturing the early stages of the disease, facilitating the potential for timely intervention. Results: The four models performed very accurately during the training phase. In fact, an accuracy higher than 99% was achieved by the ViT and ResNet models. Moreover, a lesser accuracy of 90% was achieved by the CNN five-layer model. During the test phase, only the best-performing models from the training experiments were tested. The ResNet-18 model has achieved a 100% accuracy. However, the ViT and the CNN five-layer models have achieved, respectively, 99.96% and 96.40% test accuracy. Conclusions: The results demonstrate high accuracy, highlighting the framework's capabilities, and in particular the effectiveness of the workflow used for generating the SEI images. Given the nature of the dataset used, the proposed framework stands to function as a cost-effective and accessible tool for early PD detection in various healthcare settings. This study contributes to the advancement of mobile health technologies, aiming to enhance early diagnosis and monitoring of Parkinson's Disease.

One-Dimensional Deep Residual Network with Aggregated Transformations for Internet of Things (IoT)-Enabled Human Activity Recognition in an Uncontrolled Environment

Human activity recognition (HAR) in real-world settings has gained significance due to the growth of Internet of Things (IoT) devices such as smartphones and smartwatches. Nonetheless, limitations such as fluctuating environmental conditions and intricate behavioral patterns have impacted the accuracy of the current procedures. This research introduces an innovative methodology employing a modified deep residual network, called 1D-ResNeXt, for IoT-enabled HAR in uncontrolled environments. We developed a comprehensive network that utilizes feature fusion and a multi-kernel block approach. The residual connections and the split–transform–merge technique mitigate the accuracy degradation and reduce the parameter number. We assessed our suggested model on three available datasets, mHealth, MotionSense, and Wild-SHARD, utilizing accuracy metrics, cross-entropy loss, and F1 score. The findings indicated substantial enhancements in proficiency in recognition, attaining 99.97% on mHealth, 98.77% on MotionSense, and 97.59% on Wild-SHARD, surpassing contemporary methodologies. Significantly, our model attained these outcomes with considerably fewer parameters (24,130–26,118) than other models, several of which exceeded 700,000 parameters. The 1D-ResNeXt model demonstrated outstanding effectiveness under various ambient circumstances, tackling a significant obstacle in practical HAR applications. The findings indicate that our modified deep residual network presents a viable approach for improving the dependability and usability of IoT-based HAR systems in dynamic, uncontrolled situations while preserving the computational effectiveness essential for IoT devices. The results significantly impact multiple sectors, including healthcare surveillance, intelligent residences, and customized assistive devices.

Tracking autonomic nervous system activity using surface ECG: Personalized, multiparametric evaluation

We present a concise review of the background, pitfalls, and potential solutions for the noninvasive evaluation and continuous tracking of cardiac autonomic nervous system activity (ANSA), using surface-ECG-accessible parameters, including heart rate (HR), heart-rate variability (HRV), and cardiac repolarization. These parameters have provided insights into the dynamics of cardiac ANSA in controlled experiments and have proved useful in risk assessment with respect to sudden cardiac death and all-cause mortality in some patient populations, as well as in implantable device programming. Yet attempts to translate these parameters from the laboratory environment to ambulatory settings have been hampered by the presence of multiple uncontrolled factors, including changes in blood pressure, body position, physical activity, and respiration frequency. We show that a single-parameter-based, simplified cardiac ANSA evaluation in an uncontrolled ambulatory setting could be inaccurate, and we discuss several approaches to improve accuracy. Discerning cardiac ANSA effects in uncontrolled ambulatory environments requires tracking multiple physiological processes, preferably using multisensor, multiparametric monitoring and controlling some physiological variables (e.g., respiration frequency); data fusion and machine-learning-based analytics are instrumental for developing more accurate personalized ANSA evaluation.

NU-NeRF: Neural Reconstruction of Nested Transparent Objects with Uncontrolled Capture Environment

The geometry reconstruction of transparent objects is a challenging problem due to the highly noncontinuous and rapidly changing surface color caused by refraction. Existing methods rely on special capture devices, dedicated backgrounds, or ground-truth object masks to provide more priors and reduce the ambiguity of the problem. However, it is hard to apply methods with these special requirements to real-life reconstruction tasks, like scenes captured in the wild using mobile devices. Moreover, these methods can only cope with solid and homogeneous materials, greatly limiting the scope of the application. To solve the problems above, we propose NU-NeRF to reconstruct nested transparent objects without requiring a dedicated capture environment or additional input. NU-NeRF is built upon a neural signed distance field formulation and leverages neural rendering techniques. It consists of two main stages. In Stage I, the surface color is separated into reflection and refraction. The reflection is decomposed using physically based material and rendering. The refraction is modeled using a single MLP given the refraction and view directions, which is a simple yet effective solution of refraction modeling. This step produces high-fidelity geometry of the outer surface. In stage II, we use explicit ray tracing on the reconstructed outer surface for accurate light transport simulation. The surface reconstruction is executed again inside the outer geometry to obtain any inner surface geometry. In this process, a novel transparent interface formulation is used to cope with different types of transparent surfaces. Experiments conducted on synthetic scenes and real captured scenes show that NU-NeRF is capable of producing better reconstruction results than previous methods and achieves accurate nested surface reconstruction under an uncontrolled capture environment.

Bioinspired cooperation in a heterogeneous robot swarm using ferrofluid artificial pheromones for uncontrolled environments

This article presents a novel bioinspired technology for the cooperation and coordination of heterogeneous robot swarms in uncontrolled environments, utilizing an artificial pheromone composed of magnetized ferrofluids. Communication between different types of robots is achieved indirectly through stigmergy, where messages are inherently associated with specific locations. This approach is advantageous for swarm experimentation outside controlled laboratory spaces, where localization is typically managed through centralized camera systems (e.g. infrared, RGB). Applying pheromone principles has also proven beneficial for various swarm behaviors. We introduce a detection methodology for the artificial ferrofluid pheromone using low-cost magnetic sensors, along with signal processing and parameter characterization. Experiments involved a heterogeneous swarm consisting of two types of robots: one equipped with camera and image processing capabilities and the other with basic sensor technologies. Validation in multiple uncontrolled environments (with varying floor surfaces, wind, and light conditions) demonstrated successful cooperation among robots with differing technological complexities using the proposed technology.

Improving unsupervised long-term damage detection in an uncontrolled environment through noise-augmentation strategy

Autoencoder reconstruction-based unsupervised damage detection is widely utilized in structural health monitoring. However, such methods typically necessitate a comprehensive collection of historical guided waves as training data. Acquiring such data presents challenges, as it requires prolonged monitoring to cover various environmental and operational conditions (EOCs), making these methods less practical for real-world applications. This paper proposes an unsupervised damage detection method solely trained on the current measurements directly. To improve the performance of the unsupervised damage detection method when the training data (the current measurements ) contains a large ratio of damage-induced guided waves, two noise-augmentation strategies are designed to limit the neural network’s learning ability to recover damage-induced guided waves from their segments, improving detection performance. Additionally, we use t-SNE to visualize the impact of noise augmentation on the separation of different types of guided waves within the recovery network’s latent space. Experimental results indicate that input signals with relatively low SNR can achieve better damage detection performance, and a strategy for estimating the optimal noise intensity in input signals is provided in this paper. The effectiveness of the unsupervised this damage detection method with noise-augmentation strategy is validated by 10 regions of 80-days guided waves collected from uncontrolled and dynamic environmental conditions.

Biodegrading Lignocellulosic Agricultural Waste using Phanerochaete chrysosporium and Electrical Current Stimulation

White-rot fungi (WRF), such as Phanerochaete chrysosporium, play a significant role in the lignin degradation (LD) of biomass, an essential process in the carbon recycling of terrestrial ecosystems. However, the rapid development of the agroindustry has imposed a daunting task on biomass waste management. One green initiative focuses on enhancing the bioremediation of lignin since it forms a resistant barrier to chemical and biological LD. This work demonstrated that electric current stimulation (ECS) can markedly enhance LD by P. chrysosporium. Palm oil empty fruit bunches (EFBs) were utilized as a lignin-rich substrate for P. chrysosporium. These were placed in a 250-ml enclosure filled with unbuffered potato dextrose broth (PDB) as the electrolyte. The ECS was supplied in situ in two ways: (1) by inserting a zinc anode/air electrode redox couple into the enclosure to produce a self-sustaining discharge current (DC), and (2) by connecting the enclosure to an external current (EC). The lignin content (LC) of the EFBs was assessed after 30 days of exposure to fungal microbes in an uncontrolled environment. The fungal LD rate was highest at 3 mA and even doubled under the influence of the EC, enhancing the lignin removal by 74.6%. The proposed method is much simpler and cheaper than the electrocatalytic reactions produced by the electro-Fenton method.

The rate and predictors of violence against EMS personnel

BackgroundViolence against Emergency Medical Services (EMS) personnel vary between studies. Current studies are mainly based on self-reporting, thus other designs are needed to provide more perspective. The purpose of this study was to explore the rate and predictors of violent behavior targeted at EMS personnel by exploring the Electronic patient care records (ePCR) documentation by EMS personnel.MethodsThis was a retrospective cohort study of EMS patients in Finland. The data were collected from three regions between 1st June and 30th November 2018. Text mining and manual evaluation were used to identify and explore predictors of violence targeted at EMS personnel from the ePCR narratives. Multivariable logistic regressions were used to determine factors that were independently associated with violent behavior. The results are presented with odds ratios (ORs) with 95% confidence intervals (CIs).ResultsThe EMS personnel reported experiences of violence in a total of 297 identified missions (0.7%) of all EMS missions (n = 40,263). The violence was mostly verbal (62.3%) and the most common violence perpetrator was the patient (98.0%). The police were alarmed to many missions where violence was reported (40.7%). Sometimes violence occurred suddenly although the police were present. The multivariable logistic regression model indicates that violence occurred typically in urban areas (OR 1.699; 95% CI 1.283 to 2.248), at weekend nights (OR 1.357; 95% CI 1.043 to 1.765), by male (OR 1.501; 95% CI 1.160 to 1.942), and patients influenced by alcohol (OR 3.464; 95% CI 2.644 to 4.538). A NEWS2 score of 3 in any parameter (vs. score 0–4, OR 2.386; 95% CI: 1.788 to 3.185) and ALS unit type (vs. BLS, OR 1.373; 95% CI: 1.009 to 1.866) increased the likelihood as well.ConclusionsThe documentation in ePCRs show low rates of violence targeted at EMS personnel. However, violence is a multidimensional phenomenon connected to unfamiliar patients, rushed situations, and an uncontrolled environment. This means that the EMS personnels’ safety cannot be ensured in all situations. Therefore, a balance between safety margins and treating patients needs to be considered.

Deep learning-based gamma spectroscopic analysis considering multiple variables for in situ applications

Gamma spectroscopy in environments with fluctuating temperatures, magnetic fields, and radiation doses can lead to inaccurate material analysis and increased radiation exposure for workers. In this manuscript, we propose a deep learning model that reduces human intervention and remains robust against variables in uncontrolled environments, thereby enhancing the applicability of gamma spectroscopy in the field. The process of establishing a dataset to train, validate, and test the deep learning model involved simultaneous consideration of multiple variables, including gain shifts, detector energy resolutions, the number of mixed radioisotopes (RIs), mixed RI ratios, and statistical fluctuations of the spectrum. The model was trained using multi-label learning to perform RI identification and quantification simultaneously. Specifically, the RI quantification output of our model structure was designed to predict full-energy peak areas and mixed ratios for each RI. Highlighting the importance of a high-performance training dataset, the training and validation results of the gamma spectroscopy model were analyzed in depth. In the test results, organized by RI type and number of mixed RIs, the model demonstrated that following: even under complex conditions with multiple variables, all outcomes for RI identification and quantification were predicted with high accuracy and precision, exceeding 98%.

Pain assessment from facial expression images utilizing Statistical Frei-Chen Mask (SFCM)-based features and DenseNet

Estimating pain levels is crucial for patients with serious illnesses, those recovering from brain surgery, and those receiving intensive care etc. An automatic pain intensity estimator is proposed in this study that gathers information about pain and intensity from the user’s expressions. The faces in the database are first cropped using a ‘Chehra’ face detector, which performs well even in wildly uncontrolled environments with a wide range of lighting and position fluctuations. The suggested technique extracts the beneficial and distinct patterns from facial expressions using novel Statistical Frei-Chen Mask (SFCM)-based features and DenseNet-based features. As it offers quick as well as accurate pain identification and pain intensity estimation, the Radial Basis Function Based Extreme Learning Machine (RBF-ELM) is employed for pain recognition and pain intensity level estimation using the characteristics. All the data is kept, updated and protected in the cloud because availability and high-performance decision-making are so important for informing physicians and auxiliary IoT nodes (such as wearable sensors). In addition, cloud computing reduces the time complexity of the training phase of Machine Learning algorithms in situations where it is possible to build a complete cloud/edge architecture by allocating additional computational resources and memory in use. The facial expression images from the UNBC-McMaster Shoulder Pain Expression Archive and 2D face dataset are used to test the proposed method. The measurement of pain intensity uses four stages. When compared to the results from the literature, the proposed work attains enhanced performance.

Unobtrusive measurement of cognitive load and physiological signals in uncontrolled environments

While individuals fail to assess their mental health subjectively in their day-to-day activities, the recent development of consumer-grade wearable devices has enormous potential to monitor daily workload objectively by acquiring physiological signals. Therefore, this work collected consumer-grade physiological signals from twenty-four participants, following a four-hour cognitive load elicitation paradigm with self-chosen tasks in uncontrolled environments and a four-hour mental workload elicitation paradigm in a controlled environment. The recorded dataset of approximately 315 hours consists of electroencephalography, acceleration, electrodermal activity, and photoplethysmogram data balanced across low and high load levels. Participants performed office-like tasks in the controlled environment (mental arithmetic, Stroop, N-Back, and Sudoku) with two defined difficulty levels and in the uncontrolled environments (mainly researching, programming, and writing emails). Each task label was provided by participants using two 5-point Likert scales of mental workload and stress and the pairwise NASA-TLX questionnaire. This data is suitable for developing real-time mental health assessment methods, conducting research on signal processing techniques for challenging environments, and developing personal cognitive load assistants.

Label distribution learning for compound facial expression recognition in‐the‐wild: A comparative study

AbstractHuman emotional states encompass both basic and compound facial expressions. However, current works primarily focus on basic expressions, consequently neglecting the broad spectrum of human emotions encountered in practical scenarios. Compound facial expressions involve the simultaneous manifestation of multiple emotions on an individual's face. This phenomenon reflects the complexity and richness of human states, where facial features dynamically convey a combination of feelings. This study embarks on a pioneering exploration of Compound Facial Expression Recognition (CFER), with a distinctive emphasis on leveraging the Label Distribution Learning (LDL) paradigm. This strategic application of LDL aims to address the ambiguity and complexity inherent in compound expressions, marking a significant departure from the dominant Single Label Learning (SLL) and Multi‐Label Learning (MLL) paradigms. Within this framework, we rigorously investigate the potential of LDL for a critical challenge in Facial Expression Recognition (FER): recognizing compound facial expressions in uncontrolled environments. We utilize the recently introduced RAF‐CE dataset, meticulously designed for compound expression assessment. By conducting a comprehensive comparative analysis pitting LDL against conventional SLL and MLL approaches on RAF‐CE, we aim to definitively establish LDL's superiority in handling this complex task. Furthermore, we assess the generalizability of LDL models trained on RAF‐CE by evaluating their performance on the EmotioNet and RAF‐DB Compound datasets. This demonstrates their effectiveness without domain adaptation. To solidify these findings, we conduct a comprehensive comparative analysis of 12 cutting‐edge LDL algorithms on RAF‐CE, S‐BU3DFE, and S‐JAFFE datasets, providing valuable insights into the most effective LDL techniques for FER in‐the‐wild.