3D Convolutional Neural Network Research Articles

UAV imaging hyperspectral for barnyard identification and spatial distribution in paddy fields

Barnyard grass (Echinochloa crus-galli), a noxious weed prevalent in rice fields, poses a significant threat to rice growth and development, ultimately leading to substantial yield reductions. Its morphological resemblance to rice plants complicates its identification and management. Consequently, accurately identifying and mapping barnyard grass in intricate rice field environments holds substantial research value and significance. Furthermore, most existing weed detection studies rely on single-day data, limiting the models’ generalizability. Therefore, incorporating multi-temporal data into modeling is highly desirable. This study utilized an unmanned aerial vehicle (UAV) to capture multi-temporal, low-altitude hyperspectral imagery of paddy fields, which were subsequently stitched, calibrated, and filtered using the SG convolution filter. The continuous projection algorithm (SPA) was employed to extract sensitive bands for discriminating between rice and barnyard grass. Initially, single-time data was utilized for barnyard grass identification, with models such as support vector machine (SVM), random forest (RF), 1D convolutional neural network (1DCNN), and 3D convolutional neural network (3DCNN) constructed using both full-spectrum and feature bands. The results indicated that the SPA-3DCNN model exhibited the highest performance in identifying rice (F1-score: 0.942) and barnyard grass (F1-score: 0.8936). Seven selected feature bands (482.5234, 546.5415, 675.0806, 709.1382, 762.0431, 922.0157, and 944.6371 nm) were found to be highly discriminative for distinguishing barnyard grass from rice. Subsequently, multi-temporal data modeling was employed to enhance the models’ generalization ability, resulting in improved overall accuracy (OA) and Kappa coefficients for each model, along with reduced misclassification rates. The spatial distribution maps also exhibited better performance with increased temporal data. The infestation trend of barnyard grass was visualized using multi-temporal difference maps, and a density map of barnyard grass was generated through image binarization. This study successfully demonstrated the feasibility of utilizing UAV-borne hyperspectral imagery for identifying barnyard grass in complex paddy field environments. By improving the model’s generalization ability through multi-temporal modeling, the study provided robust data support for managing and preventing barnyard grass infestations, including the creation of spatial distribution and density maps for barnyard grass.

Enhancing bridge damage assessment: Adaptive cell and deep learning approaches in time-series analysis

In recent years, the application of Deep Learning (DL) for damage detection in Structural Health Monitoring (SHM) using time-series data has garnered significant attention from the global scientific community. By incorporating numerous filters and units in their architecture, these models facilitate feature extraction making the processing of large time-series datasets in SHM more effective. However, optimizing of the number of filters and units is a challenge, often relying on scientists' experience or previous models without a systematic method for measurement or automatic selection. Therefore, this study proposes a novel framework to enhance the accuracy and efficiency of SHM for damage detection in bridge structures by using metaheuristic algorithms, Electric Eel Foraging Optimization (EEFO) algorithm, to optimize hyperparameters of DL model. This DL model combines the advantages of each traditional model. Specifically, 1D Convolutional Neural Network (1DCNN) is employed for features extraction, Gated Recurrent Units (GRU) for identifying long-term dependencies, and Residual Networks (ResNet) for avoiding vanishing gradient problem which often happens in DL model during training, referred to as 1DCNN-GRU-ResNet. The time-series dataset generated from Cua Rao bridge is used to demonstrate the effectiveness of the proposed method. 1DCNN-GRU-ResNet after optimization (1DCNN-GRU-ResNet-opt) achieves 91.6 %, significantly surpassing traditional 1DCNN (82.5 %), GRU (79.9 %), 1DCNN-GRU (85.7 %), and 1DCNN-GRU-ResNet (89.3 %) in the test set. The proposed 1DCNN-GRU-ResNet-opt approach demonstrates considerable potential in practical applications for SHM, offering high accuracy and efficiency.

Artificial intelligence detects awareness of functional relation with the environment in 3 month old babies

A recent experiment probed how purposeful action emerges in early life by manipulating infants’ functional connection to an object in the environment (i.e., tethering an infant’s foot to a colorful mobile). Vicon motion capture data from multiple infant joints were used here to create Histograms of Joint Displacements (HJDs) to generate pose-based descriptors for 3D infant spatial trajectories. Using HJDs as inputs, machine and deep learning systems were tasked with classifying the experimental state from which snippets of movement data were sampled. The architectures tested included k-Nearest Neighbour (kNN), Linear Discriminant Analysis (LDA), Fully connected network (FCNet), 1D-Convolutional Neural Network (1D-Conv), 1D-Capsule Network (1D-CapsNet), 2D-Conv and 2D-CapsNet. Sliding window scenarios were used for temporal analysis to search for topological changes in infant movement related to functional context. kNN and LDA achieved higher classification accuracy with single joint features, while deep learning approaches, particularly 2D-CapsNet, achieved higher accuracy on full-body features. For each AI architecture tested, measures of foot activity displayed the most distinct and coherent pattern alterations across different experimental stages (reflected in the highest classification accuracy rate), indicating that interaction with the world impacts the infant behaviour most at the site of organism~world connection.

One-dimensional deep learning driven geospatial analysis for flash flood susceptibility mapping: a case study in North Central Vietnam

Flash floods rank among the most catastrophic natural disasters worldwide, inflicting severe socio-economic, environmental, and human impacts. Consequently, accurately identifying areas at potential risk is of paramount importance. This study investigates the efficacy of Deep 1D-Convolutional Neural Networks (Deep 1D-CNN) in spatially predicting flash floods, with a specific focus on the frequent tropical cyclone-induced flash floods in Thanh Hoa province, North Central Vietnam. The Deep 1D-CNN was structured with four convolutional layers, two pooling layers, one flattened layer, and two fully connected layers, employing the ADAM algorithm for optimization and Mean Squared Error (MSE) for loss calculation. A geodatabase containing 2540 flash flood locations and 12 influencing factors was compiled using multi-source geospatial data. The database was used to train and check the model. The results indicate that the Deep 1D-CNN model achieved high predictive accuracy (90.2%), along with a Kappa value of 0.804 and an AUC (Area Under the Curve) of 0.969, surpassing the benchmark models such as SVM (Support Vector Machine) and LR (Logistic Regression). The study concludes that the Deep 1D-CNN model is a highly effective tool for modeling flash floods.

Eco-friendly mix design of slag-ash-based geopolymer concrete using explainable deep learning

Geopolymer concrete is a sustainable and eco-friendly substitute for traditional OPC (Ordinary Portland Cement) based concrete, as it reduces greenhouse gas emissions. With various supplementary cementitious materials, the compressive strength of geopolymer concrete should be accurately predicted. Recent studies have applied deep learning techniques to predict the compressive strength of geopolymer concrete yet its hidden decision-making criteria diminish the end-users’ trust in predictions. To bridge this gap, the authors first developed three deep learning models: an artificial neural network (ANN), a deep neural network (DNN), and a 1D convolution neural network (CNN) to predict the compressive strength of slag ash-based geopolymer concrete. The performance indices for accuracy revealed that the DNN model outperforms the other two models. Subsequently, Shapley additive explanations (SHAP) were used to explain the best-performed deep learning model, DNN, and its compressive strength predictions. SHAP exhibited how the importance of each feature and its relationship contributes to the compressive strength prediction of the DNN model. Finally, the authors developed a novel DNN-based open-source software interface to predict the mix design proportions for a given target compressive strength (using inverse modeling technique) for slag ash-based geopolymer concrete. Additionally, the software calculates the Global Warming Potential (kg CO2 equivalent) for each mix design to select the mix designs with low greenhouse emissions.

Subject-Independent Model for Reconstructing Electrocardiography Signals from Photoplethysmography Signals

Electrocardiography (ECG) is the gold standard for monitoring vital signs and for diagnosing, controlling, and preventing cardiovascular diseases (CVDs). However, ECG requires continuous user participation, and cannot be used for continuous cardiac monitoring. In contrast to ECG, photoplethysmography (PPG) devices do not require continued user involvement, and can offer ongoing and long-term detection capabilities. However, from a medical perspective, ECG can provide more information about the heart. Currently, most existing work contains different signals recorded from the same subject in training and test sets. This study proposes a neural network model based on a 1D convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) network. This neural network model can directly reconstruct ECG signals from PPG signals. The learned features are captured from the CNN model and fed into the BiLSTM model. In order to verify the validity of the model, it is evaluated using the MIMIC II dataset in the completely subject-independent model (records are placed in a training set, and a test set appears once, but the test signal belongs to a record that is not in the training set). The Pearson’s correlation coefficient between the reconstructed ECG and the reference ECG of the proposed model is 0.963 in the completely subject-independence model. The results of the proposed model are better than those of several cited state-of-the-art models. The results of our trained model indicate that we can obtain reconstructed ECGs that are highly similar to reference ECGs in the completely subject-independent model.

Comparison of deep learning architectures for predicting amyloid positivity in Alzheimer's disease, mild cognitive impairment, and healthy aging, from T1-weighted brain structural MRI.

Abnormal β-amyloid (Aβ) accumulation in the brain is an early indicator of Alzheimer's disease (AD) and is typically assessed through invasive procedures such as PET (positron emission tomography) or CSF (cerebrospinal fluid) assays. As new anti-Alzheimer's treatments can now successfully target amyloid pathology, there is a growing interest in predicting Aβ positivity (Aβ+) from less invasive, more widely available types of brain scans, such as T1-weighted (T1w) MRI. Here we compare multiple approaches to infer Aβ + from standard anatomical MRI: (1) classical machine learning algorithms, including logistic regression, XGBoost, and shallow artificial neural networks, (2) deep learning models based on 2D and 3D convolutional neural networks (CNNs), (3) a hybrid ANN-CNN, combining the strengths of shallow and deep neural networks, (4) transfer learning models based on CNNs, and (5) 3D Vision Transformers. All models were trained on paired MRI/PET data from 1,847 elderly participants (mean age: 75.1 yrs. ± 7.6SD; 863 females/984 males; 661 healthy controls, 889 with mild cognitive impairment (MCI), and 297 with Dementia), scanned as part of the Alzheimer's Disease Neuroimaging Initiative. We evaluated each model's balanced accuracy and F1 scores. While further tests on more diverse data are warranted, deep learning models trained on standard MRI showed promise for estimating Aβ + status, at least in people with MCI. This may offer a potential screening option before resorting to more invasive procedures.

Automatic segmentation of the spinal cord nerve rootlets

Abstract Precise identification of spinal nerve rootlets is relevant to delineate spinal levels for the study of functional activity in the spinal cord. The goal of this study was to develop an automatic method for the semantic segmentation of spinal nerve rootlets from T2-weighted magnetic resonance imaging (MRI) scans. Images from two open-access 3T MRI datasets were used to train a 3D multi-class convolutional neural network using an active learning approach to segment C2-C8 dorsal nerve rootlets. Each output class corresponds to a spinal level. The method was tested on 3T T2-weighted images from three datasets unseen during training to assess inter-site, inter-session, and inter-resolution variability. The test Dice score was 0.67 ± 0.16 (mean ± standard deviation across testing images and rootlets levels), suggesting a good performance. The method also demonstrated low inter-vendor and inter-site variability (coefficient of variation ≤ 1.41%), as well as low inter-session variability (coefficient of variation ≤ 1.30%), indicating stable predictions across different MRI vendors, sites, and sessions. The proposed methodology is open-source and readily available in the Spinal Cord Toolbox (SCT) v6.2 and higher.

Application of multi-label convolutional neural networks on enhancement of reliability of classification, a case study on bearing fault diagnostics

In contrast to the mutli-class classification where one and only one class would be assigned to each instance, multi-label classification allows for more flexible classification and possibility of having multiple labels for instance. This property enables us to model the dependencies between the classes such as effect of one label on the probability of occurrence of the other label. Considering these connections between the classes can help us to enhance the reliability of classifier and handle the contradictory prediction. Much like how we can assign multiple classes to an instance, we can also have an instance that remains unlabeled. This class is typically referred to as the "neutral" label, allowing us to bluntly state that we lack sufficient information to label the instance with adequate certainty. Leveraging these properties, a 1D convolutional neural network (CNN) has been proposed to classify common faults occurring in the bearings. Network has the capability to assign a Confidence level between 0 to 1 to each class independently. Later, a threshold has been set on each class to identify the instance labels if any. Subsequently, instances with contradictory predicted label have been flagged as undetermined and removed from the predicted result. thus, the remaining classified instances can be labeled with sufficient confident. In the end, the multi-label CNN model performance has been compared with equivalent multi-class CNN model. Technique has been evaluated using MAFAULDA open dataset which is composed of 1951 multivariate time-series acquired by a triaxial accelerometer, a microphone, and a tachometer using SpectraQuest’s Machinery Fault Simulator (MFS) experimentally simulating different bearing states including normal operation as well as inner and outer race bearing faults. A data reduction scheme has been performed on the dataset to make the diagnostic more challenging and evaluate the performance of CNN in more realistic condition with insufficient data.

A Novel Supercapacitor Degradation Prediction Using a 1D Convolutional Neural Network and Improved Informer Model

A Novel Supercapacitor Degradation Prediction Using a 1D Convolutional Neural Network and Improved Informer Model

A deep transfer learning model for the deformation of braced excavations with limited monitoring data

The current deep learning models for braced excavation cannot predict deformation from the beginning of excavation due to the need for a substantial corpus of sufficient historical data for training purposes. To address this issue, this study proposes a transfer learning model based on a sequence-to-sequence two-dimensional (2D) convolutional long short-term memory neural network (S2SCL2D). The model can use the existing data from other adjacent similar excavations to achieve wall deflection prediction once a limited amount of monitoring data from the target excavation has been recorded. In the absence of adjacent excavation data, numerical simulation data from the target project can be employed instead. A weight update strategy is proposed to improve the prediction accuracy by integrating the stochastic gradient masking with an early stopping mechanism. To illustrate the proposed methodology, an excavation project in Hangzhou, China is adopted. The proposed deep transfer learning model, which uses either adjacent excavation data or numerical simulation data as the source domain, shows a significant improvement in performance when compared to the non-transfer learning model. Using the simulation data from the target project even leads to better prediction performance than using the actual monitoring data from other adjacent excavations. The results demonstrate that the proposed model can reasonably predict the deformation with limited data from the target project.

Fdstnet & TransCAM: A 1D CNN temperature interpretation model for temperature indication paint and model interpretability method

The color of temperature indicating paint (TIP) is commonly used in temperature measurement testing of aircraft engine hot end components. In this paper, an end-to-end convolutional neural network utilizing DSTNet, is proposed for accurate temperature interpretation of TIP. Furthermore, the DSTNet is optimized by fusion of reflectance spectrum and sub-band statistical features, resulting in the creation of FDSTNet. The FDSTNet achieves highly accurate temperature interpretation. When the temperature error is 0℃, the interpretation accuracy on KN3A, KN6 and KN8 samples is 98 %, 97 % and 94 %, respectively. When the temperature error is ±10℃, the accuracy on the engine blade is 88 %. In addition, a 1D convolutional neural network interpretability method based on TransCAM, is proposed. This approach analyzes the process of learning the reflectance spectrum of DSTNet, and localizes the characteristic wavelength by visualizing the class activation mapping curves of each convolutional layer of DSTNet.

Quantification of cauda equina nerve root dispersion through radiomics features in weight-bearing MRI in normal subjects and spinal canal stenosis patients.

To quantify the distribution of cauda equina nerve roots in supine and upright positions using manual measurements and radiomics features both in normal subjects and in lumbar spinal canal stenosis (LSCS) patients. We retrospectively recruited patients who underwent weight-bearing MRI in supine and upright positions for back pain. 3D T2-weighted isotropic acquisition (3D-HYCE) sequences were used to develop a 3D convolutional neural network for identification and segmentation of lumbar vertebrae. Para-axial reformatted images perpendicular to the spinal canal and parallel to each vertebral endplate were automatically extracted. From each level, we computed the maximum antero-posterior (AP) and latero-lateral (LL) dispersion of nerve roots; further, radiomics features were extracted to quantify standardized metrics of nerve root distribution. We included 16 patients with LSCS and 20 normal subjects. In normal subjects, nerve root AP dispersion significantly increased from supine to upright position (p < 0.001, L2-L5 levels), and radiomics features showed an increase in non-uniformity. In LSCS subjects, in the upright position AP dispersion of nerve roots and entropy-related features increased caudally to the stenosis level (p < 0.001) and decreased cranially (p < 0.001). Moreover, entropy-related radiomics features negatively correlated with pre-operative Pain Numerical Rating Scale. Comparison between normal subjects and LSCS patients showed a difference in AP dispersion and increase of variance cranially to the stenosis level (p < 0.001) in the upright position. Nerve root distribution inside the dural sac changed between supine and upright positions, and radiomics features were able to quantify the differences between normal and LSCS subjects. The distribution of cauda equina nerve roots and the redundant nerve root sign significantly varies between supine and upright positions in normal subjects and spinal canal stenosis patients, respectively. Radiomics features quantify nerve root dispersion and correlates with pain severity. • Weight-bearing MRI depicts spatial distribution of the cauda equina in both supine and upright positions in normal subjects and spinal stenosis patients. • Radiomics features can quantify the effects of spinal stenosis on the dispersion of the cauda equina in the dural sac. • In the orthostatic position, dispersion of nerve roots is different in lumbar spinal stenosis patients compared to that in normal subjects; entropy-related features negatively correlated with pre-operative Pain Numerical Rating Scale.

Network traffic prediction by learning time series as images

Network traffic prediction is crucial for cost-effective network management, resource allocation, and security in the emergent software-defined and zero-touch networks. Machine Learning (ML) models such as Long-Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) have shown great promise in traffic prediction. This work proposes a hybrid ML model consisting of pre-trained AlexNet and ResNet models, the 2D Convolutional Neural Network (CNN), Vision Transformer (ViT), and LSTM/GRU models and their variants to predict network traffic. This work explores two approaches for converting the time series data to images by allowing more precise feature extraction and then performing traffic prediction on an image dataset, thus increasing accuracy. Given their proficiency in extracting image features we use three pre-trained models: the AlexNet, the ResNet combined with CNN, and the ViT. Then, we use those models as inputs to the LSTM, Bi-LSTM, and the GRU, Bi-GRU models for predicting network traffic data. The models are evaluated on the GÉANT and Abilene traffic networks and assessed based on the Mean Square Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Comparison of results with existing network traffic prediction techniques shows considerable improvements in all performance metrics revealed by the Wilcoxon test. Our best model results for the GÉANT and Abilene datasets are respectively; MSE of 0.00058 and 0.03078, RMSE of 0.02415 and 0.17544, and MAE of 0.00774 and 0.13384.

Construction of a network intrusion detection system based on a convolutional neural network and a bidirectional gated recurrent unit with attention mechanism

The object of this study is the process of recognizing intrusions in computer networks. Network intrusion detection systems (NIDS) have become an important area of research as they are used to protect computer systems from hacker attacks. Deep learning is becoming increasingly popular for detecting and classifying malicious network traffic, including for building NIDS. In this paper, we propose a network intrusion detection model CNN-BiGRU-Attention based on a time-based approach to deep learning using the attention mechanism. The main goal of the study is to build an effective combined deep learning model that can detect various network cyber threats. A 1D convolutional neural network is implemented to extract high-level representations of intrusion information features. A bidirectional gated recurrent unit (BiGRU) with an attention mechanism for traffic data classification has been designed. The attention mechanism plays a key role in the constructed model as it allows the system to focus only on important aspects of network traffic and allows the model to adapt to new types of threats. The results of the study show that using a combination of CNN and BiGRU with the attention mechanism speeds up and improves the process of classifying network attacks. On the NSL-KDD and UNSW-NB15 training datasets, the model shows an accuracy of 99.81 % and 97.80 %. On the NSL-KDD and UNSW-NB15 test datasets, the model demonstrates 82.16 % and 97.72 % accuracy. The proposed NIDS model will be considered for implementation in a real-time corporate network security system. In general, the results of the study provide a new perspective on improving the performance of NIDS and are quite relevant in terms of using attention mechanisms to classify network traffic

AI-based visual speech recognition towards realistic avatars and lip-reading applications in the metaverse

The metaverse, a virtually shared digital world where individuals interact, create, and explore, has witnessed rapid evolution and widespread adoption. Communication between avatars is crucial to their actions in the metaverse. Advances in natural language processing have allowed for significant progress in producing spoken conversations. Within this digital landscape, the integration of Visual Speech Recognition (VSR) powered by deep learning emerges as a transformative application. This research delves into the concept and implications of VSR in the metaverse. This study focuses on developing realistic avatars and a lip-reading application within the metaverse, utilizing Artificial Intelligence (AI) techniques for visual speech recognition. Visual Speech Recognition in the metaverse refers to using deep learning techniques to comprehend and respond to spoken language, relying on the visual cues provided by users' avatars. This multidisciplinary approach combines computer vision and natural language processing, enabling avatars to understand spoken words by analyzing the movements of their lips and facial expressions. Key components encompass the collection of extensive video datasets, the employment of 3D Convolutional Neural Networks (3D CNNs) combined with ShuffleNet and Densely Connected Temporal Convolutional Neural Networks (DC-TCN) called (CFS-DCTCN) to model visual and temporal features, and the integration of contextual understanding mechanisms. The two datasets Wild (LRW) dataset and the GRID Corpus datasets are utilized to validate the proposed model. As the metaverse continues its prominence, integrating Visual Speech Recognition through deep learning represents a pivotal step towards forging immersive and dynamic virtual worlds where communication transcends physical boundaries. This paper contributes to the foundation of technology-driven metaverse development and fosters a future where digital interactions mirror the complexities of human communication. The proposed model achieves 99.5 % on LRW and 98.8 % on the GRID dataset.

Higher-Order Spectral Analysis Combined with a Convolution Neural Network for Atrial Fibrillation Detection-Preliminary Study.

The global burden of atrial fibrillation (AFIB) is constantly increasing, and its early detection is still a challenge for public health and motivates researchers to improve methods for automatic AFIB prediction and management. This work proposes higher-order spectra analysis, especially the bispectrum of electrocardiogram (ECG) signals combined with the convolution neural network (CNN) for AFIB detection. Like other biomedical signals, ECG is non-stationary, non-linear, and non-Gaussian in nature, so the spectra of higher-order cumulants, in this case, bispectra, preserve valuable features. The two-dimensional (2D) bispectrum images were applied as input for the two CNN architectures with the output AFIB vs. no-AFIB: the pre-trained modified GoogLeNet and the proposed CNN called AFIB-NET. The MIT-BIH Atrial Fibrillation Database (AFDB) was used to evaluate the performance of the proposed methodology. AFIB-NET detected atrial fibrillation with a sensitivity of 95.3%, a specificity of 93.7%, and an area under the receiver operating characteristic (ROC) of 98.3%, while for GoogLeNet results for sensitivity and specificity were equal to 96.7%, 82%, respectively, and the area under ROC was equal to 96.7%. According to preliminary studies, bispectrum images as input to 2D CNN can be successfully used for AFIB rhythm detection.

AI-powered sensor fault detection for cost-effective smart greenhouses

Sensor networks in greenhouses play a pivotal role in controlling the stability of environmental and chemical factors. The Internet of Things (IoT) has been widely adopted for monitoring various sensor networks in greenhouses. In the present study, an IoT platform for remote monitoring of greenhouse environment is designed. The platform consists of a sensor node, a sink node, and an edge server. The sensor node measures indoor and outdoor humidity and temperature, interior CO(g), and interior luminosity and transmits data to the sink node, where it is timestamped. The sink node is connected to an edge server through the Mosquitto MQTT broker and the data is subsequently transferred to a MongoDB Cloud infrastructure, where the data of each variable is stored in proper formats. In the second part of the study, four 1D convolutional neural networks (CNNs) were developed for data prediction of each sensor to provide fault-tolerance in the system. The first three models, for predicting inner humidity and temperature, outdoor temperature, and outdoor humidity, directly predict the actual data of the faulty sensor based on regression analysis. The last model is designed for predicting CO and luminosity and performs data classification for faulty sensors. The models provide a high level of precision in their predictions. The RMSE for interior temperature and humidity and exterior temperature and humidity are 0.86 ℃, 3.47%, 0.682 ℃, and 2.74%, respectively. Additionally, the accuracy for luminosity and CO classification are 89.70% and 83.43%. Comparison of 1D CNN, decision tree, and linear regression model revealed that the machine learning models considerably outperform linear regression model, and they can better capture the nonlinear correlations among the variables. Furthermore, the predictive outcomes of the models were consistent across different weather conditions. The proposed methodology can be used to induce tolerance against faulty reads at sensor level in greenhouse sensor networks, independent of time and the data gathered by the faulty sensors. The study indicates that exploiting cloud resources for promoting the use of complex AI models on IoT platforms can provide a suitable solution for real-time monitoring and fault control of greenhouse environmental factors.

Deep Learning for Flash Drought Detection: A Case Study in Northeastern Brazil

Flash droughts (FDs) pose significant challenges for accurate detection due to their short duration. Conventional drought monitoring methods have difficultly capturing this rapidly intensifying phenomenon accurately. Machine learning models are increasingly useful for detecting droughts after training the models with data. Northeastern Brazil (NEB) has been a hot spot for FD events with significant ecological damage in recent years. This research introduces a novel 2D convolutional neural network (CNN) designed to identify spatial FDs in historical simulations based on multiple environmental factors and thresholds as inputs. Our model, trained with hydro-climatic data, provides a probabilistic drought detection map across northeastern Brazil (NEB) in 2012 as its output. Additionally, we examine future changes in FDs using the Coupled Model Intercomparison Project Phase 6 (CMIP6) driven by outputs from Shared Socioeconomic Pathways (SSPs) under the SSP5-8.5 scenario of 2024–2050. Our results demonstrate that the proposed spatial FD-detecting model based on 2D CNN architecture and the methodology for robust learning show promise for regional comprehensive FD monitoring. Finally, considerable spatial variability of FDs across NEB was observed during 2012 and 2024–2050, which was particularly evident in the São Francisco River Basin. This research significantly contributes to advancing our understanding of flash droughts, offering critical insights for informed water resource management and bolstering resilience against the impacts of flash droughts.

Snow-CLOCs: Camera-LiDAR Object Candidate Fusion for 3D Object Detection in Snowy Conditions.

Although existing 3D object-detection methods have achieved promising results on conventional datasets, it is still challenging to detect objects in data collected under adverse weather conditions. Data distortion from LiDAR and cameras in such conditions leads to poor performance of traditional single-sensor detection methods. Multi-modal data-fusion methods struggle with data distortion and low alignment accuracy, making accurate target detection difficult. To address this, we propose a multi-modal object-detection algorithm, Snow-CLOCs, specifically for snowy conditions. In image detection, we improved the YOLOv5 algorithm by integrating the InceptionNeXt network to enhance feature extraction and using the Wise-IoU algorithm to reduce dependency on high-quality data. For LiDAR point-cloud detection, we built upon the SECOND algorithm and employed the DROR filter to remove noise, enhancing detection accuracy. We combined the detection results from the camera and LiDAR into a unified detection set, represented using a sparse tensor, and extracted features through a 2D convolutional neural network to achieve object detection and localization. Snow-CLOCs achieved a detection accuracy of 86.61% for vehicle detection in snowy conditions.