High Classification Accuracy Research Articles

Merging holography, fluorescence, and machine learning for in situ continuous characterization and classification of airborne microplastics

Abstract. The continued increase in global plastic production and poor waste management ensures that plastic pollution will be a serious environmental concern for years to come. Because of their size, shape, and relatively low density, plastic particles between 1 and 1000 µm in size (known as microplastics or MPs) emitted directly into the environment (“primary”) or created due to degradation (“secondary”) may be transported through the atmosphere, similarly to other coarse-mode particles such as mineral dust. MPs can thus be advected over great distances, reaching even the most pristine and remote areas of Earth, and may have significant negative consequences for humans and the environment. The detection and analysis of MPs once airborne, however, remains a challenge because most observational methods are offline and resource-intensive and, therefore, not capable of providing continuous quantitative information. In this study, we present results using an online in situ airflow cytometer (SwisensPoleno Jupiter; Swisens AG; Emmen, Switzerland) – coupled with machine learning – to detect, analyze, and classify airborne single-particle MPs in near real time. The performance of the instrument in differentiating between single-particle MPs of five common polymer types (including polypropylene, polyethylene, polyamide, poly(methyl methacrylate), and polyethylene terephthalate) was investigated under laboratory conditions using combined information about their size and shape (determined using holographic imaging) and fluorescence measured using three excitation wavelengths and five emission detection windows. The classification capability using these methods was determined alongside other coarse-mode aerosol particles with similar morphology or fluorescence characteristics, such as a mineral dust and several pollen taxa. The tested MPs exhibit a measurable fluorescence signal that not only allows them to be distinguished from other fluorescent particles, such as pollen, but also differentiated from each other, with high (> 90 %) classification accuracy based on their multispectral fluorescence signatures. The classification accuracies of machine learning models using only holographic images of particles, only the fluorescence response, and combined information from holography and fluorescence to predict particle types are presented and compared. The last model, using both the holographic images and fluorescence information for each particle, was the most optimal model used, providing the highest classification accuracy compared to employing models using only the holography or fluorescence response separately. The results provide a foundation for significantly improving the understanding of the properties and types of MPs present in the atmosphere.

Method to improve the classification accuracy by in situ laser cleaning of painted metal scraps during laser-induced breakdown spectroscopy based sorting

Abstract Scrap metals are typically covered with surface contaminants, such as paint, dust, and rust, which can significantly affect the emission spectrum during laser-induced breakdown spectroscopy (LIBS) based sorting. In this study, the effects of paint layers on metal surfaces during LIBS classification were investigated. LIBS spectra were collected from metal surfaces painted with black and white paints by ablation with a nanosecond pulsed laser (wavelength = 1064 nm, pulse width = 7 ns). For the black-painted samples, the LIBS spectra showed a broad background emission, emission lines unrelated to the target metals, large shot-to-shot variation, and a relatively low signal intensity of the target metal, causing poor classification accuracy even at high shot numbers. Cleaning the black paint layer by ablating over a wide area prior to LIBS analysis resulted in high classification accuracy with fewer shot numbers. A method to determine the number of cleaning shots necessary to obtain high classification accuracy and high throughput is proposed on the basis of the change in LIBS signal intensity during cleaning shots. For the white-painted samples, the paint peeled off the metal surface after the first shot, and strong LIBS signals were measured after the following shot, which were attributed to the nanoparticles generated by the ablation of the paint, allowing an accurate classification after only two shots. The results demonstrate that different approaches must be employed depending on the paint color to achieve high classification accuracy with fewer shot numbers.

Privacy-preserving detection and classification of diabetic retinopathy using federated learning with FedDEO optimization

Diabetic retinopathy (DR) is a major cause of blindness among adults worldwide. Detecting and classifying DR early is essential for timely treatment and prevention of vision loss. This study introduces a new approach to identify and classify DR by using federated learning (FL) environment and Federated differential evolution optimization (FedDEO). FL enables collaborative learning across multiple decentralized devices while maintaining data privacy. FedDEO optimization enhances the model's performance by fine-tuning hyperparameters in a distributed manner. The research achieves state-of-the-art accuracy in DR classification using the MESSIDOR dataset for training and testing. The findings suggest that integrating FedDEO optimization within the FL framework can significantly improve DR detection, offering a scalable and privacy-preserving solution for ophthalmic healthcare. The FedDEO algorithm optimizes hyperparameters, including learning rates and batch sizes, to enhance model performance. The experiments show that FedDEO works effectively for the MESSIDOR dataset to achieve a high classification accuracy, specificity, recall, and F1-Measure of 96.98%, 98.12%, 97.12%, and 98.00% while preserving privacy. The results demonstrate that implementing these algorithms in an FL environment significantly improves privacy and performance compared to other approaches. Specifically, the proposed technique outperforms others in terms of accuracy, specificity, recall, and F1 score when applied to categorizing retinal images.

Diffusion‐Based Causality‐Preserving Neural Network for Dementia Recognition

ABSTRACTAnalyzing large‐scale functional brain networks for brain disorders often relies on undirected correlations in activation signals between brain regions. While focusing on co‐occurring activations, this approach overlooks the potential for directionality inherent in brain connectivity. Established research indicates the causal nature of brain networks, suggesting that activation patterns co‐occur and potentially influence one another. To this end, we propose a novel dffusion vector auto‐regressive (Diff‐VAR) method, enabling the assessment of whole‐brain effective connectivity (EC) as a directed and weighted network by integrating the search objectives into the deep neural network model as learnable parameters. The EC learned by our method identifies widespread differences in flow of influence within the brain network for individuals with impaired brain function compared to those with normal brain function. Moreover, we introduce an adaptive smoothing mechanism to enhance the stability and reliability of inferred EC. We evaluated the results of our proposed method on the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The model's performance is compared with existing correlation‐based and causality‐based methods. The results revealed that the brain networks constructed by our method achieve high classification accuracy and exhibit features consistent with physiological mechanisms. The code is available at https://github.com/SaqibMamoon/Diff‐VAR.

A comparison of two dynamic assessment situations for detecting development language disorder in monolingual and bilingual children

ABSTRACT Bilingual children’s language skills are strongly influenced by exposure to each of their languages, among other linguistic, environmental, and cognitive factors. In the speech and language therapy clinic, it is difficult to disentangle developmental language disorders from insufficient exposure. Dynamic assessment, which directly tests the learning potential of children, offers a promising solution for this dilemma. This study compares the clinical potential of two dynamic assessment situations, varying amount of adult mediation (autonomous computer game vs. interactive story reading with graduated cues), as well as item types (nouns, verbs, and inflections in sentences) and linguistic modalities and tasks (comprehension – word picture matching and acceptability judgement, production – free recall and picture naming). Forty-nine French monolingual and French-Portuguese bilingual children with and without developmental language disorder, aged 5;0 to 7;11 years, were include in the final analyses. Using Lasso regressions, we were able to determine which variables best explain the presence of disorder. A combination of all item types and predominantly receptive tasks, mostly from the interactive situation, was retained for very high classification accuracy (up to 100% sensitivity and 96% specificity). Language status showed no influence, which encourages the use of dynamic assessment in the context of speech and language assessment with bilingual children. This study adds to evidence that dynamic assessment is a promising task for identifying bilingual and monolingual children with developmental language disorder, particularly when the situation involves interaction with graduated cues.

Plant Disease Classification

Plant diseases significantly impact agricultural productivity, leading to economic losses and food insecurity worldwide. Timely and accurate detection of plant diseases is critical to mitigate these effects and ensure sustainable farming practices. This project explores the use of deep learning techniques for detecting plant diseases based on leaf images. The system leverages convolutional neural networks (CNNs), specifically pre-trained models like ResNet and MobileNet, fine-tuned on the publicly available PlantVillage dataset. The dataset consists of thousands of labeled images of healthy and diseased leaves from various crops. To enhance model performance and generalization, data augmentation techniques such as rotation, flipping, and brightness adjustments were applied during preprocessing. The proposed system achieves high classification accuracy, validated using metrics such as precision, recall, and F1-score. Additionally, visualization tools like Grad-CAM are used to interpret model predictions, highlighting regions of the leaf that influence the decision-making process. The model is further optimized for deployment on mobile and web platforms, enabling real-time disease diagnosis. This approach offers an efficient, scalable, and user-friendly solution for farmers and agricultural experts, aiding in early disease detection and contributing to improved crop management and yield. Future work involves expanding the dataset, incorporating more plant species, and integrating the model with IoT devices for field application

Sentiment Analysis and Rating Predicting for Hotel Review

This paper provides a comprehensive system for sentiment analysis and rate prediction of hotel reviews, combining modern natural language processing techniques with machine learning algorithms. The major purpose is to automatically assess consumer input and forecast related scores based on textual reviews. The system employs regression models to predict star ratings from the textual data and sentiment analysis to categorise reviews into good, negative, or neutral categories. The dataset consists of a large corpus of hotel reviews collected from various online platforms. Experimental results demonstrate the effectiveness of the proposed approach, with high accuracy in sentiment classification and rating prediction. The findings indicate that automated sentiment analysis and rating prediction can provide valuable insights for hotel management, helping to enhance customer satisfaction and improve sequence.

Evaluating deep learning models for classifying OCT images with limited data and noisy labels

The use of deep learning for OCT image classification could enhance the diagnosis and monitoring of retinal diseases. However, challenges like variability in retinal abnormalities, noise, and artifacts in OCT images limit its clinical use. Our study aimed to evaluate the performance of various deep learning (DL) architectures in classifying retinal pathologies versus healthy cases based on OCT images, under data scarcity and label noise. We examined five DL architectures: ResNet18, ResNet34, ResNet50, VGG16, and InceptionV3. Fine-tuning of the pre-trained models was conducted on 5526 OCT images and reduced subsets down to 21 images to evaluate performance under data scarcity. The performance of models fine-tuned on subsets with label noise levels of 10%, 15%, and 20% was evaluated. All DL architectures achieved high classification accuracy (> 90%) with training sets of 345 or more images. InceptionV3 achieved the highest classification accuracy (99%) when trained on the entire training set. However, classification accuracy decreased and variability increased as sample size decreased. Label noise significantly affected model accuracy. Compensating for labeling errors of 10%, 15%, and 20% requires approximately 4, 9, and 14 times more images in the training set to reach the performance of 345 correctly labeled images. The results showed that DL models fine-tuned on sets of 345 or more OCT images can accurately classify retinal pathologies versus healthy controls. Our findings highlight that while mislabeling errors significantly impact classification performance in OCT analysis, this can be effectively mitigated by increasing the training sample size. By addressing data scarcity and labeling errors, our research aims to improve the real-world application and accuracy of retinal disease management.

Artificial intelligence enabled smart mask for speech recognition for future hearing devices

In recent years, Lip-reading has emerged as a significant research challenge. The aim is to recognise speech by analysing Lip movements. The majority of Lip-reading technologies are based on cameras and wearable devices. However, these technologies have well-known occlusion and ambient lighting limitations, privacy concerns as well as wearable device discomfort for subjects and disturb their daily routines. Furthermore, in the era of coronavirus (COVID-19), where face masks are the norm, vision-based and wearable-based technologies for hearing aids are ineffective. To address the fundamental limitations of camera-based and wearable-based systems, this paper proposes a Radio Frequency Identification (RFID)-based smart mask for a Lip-reading framework capable of reading Lips under face masks, enabling effective speech recognition and fostering conversational accessibility for individuals with hearing impairment. The system uses RFID technology to make Radio Frequency (RF) sensing-based Lip-reading possible. A smart RFID face mask is used to collect a dataset containing three different classes of vowels (A, E, I, O, U), Consonants (F, G, M, S), and words (Fish, Goat, Meal, Moon, Snake). The collected data are fed into well-known machine-learning models for classification. A high classification accuracy is achieved by individual classes and combined datasets. On the RFID combined dataset, the Random Forest model achieves a high classification accuracy of 80%.

Investigating the Triple Code Model in numerical cognition using stereotactic electroencephalography.

The ability to conceptualize numerical quantities is an essential human trait. According to the "Triple Code Model" in numerical cognition, distinct neural substrates encode the processing of visual, auditory, and non-symbolic numerical representations. While our contemporary understanding of human number cognition has benefited greatly from advances in clinical imaging, limited studies have investigated the intracranial electrophysiological correlates of number processing. In this study, 13 subjects undergoing stereotactic electroencephalography for epilepsy participated in a number recognition task. Drawing upon postulates of the Triple Code Model, we presented subjects with numerical stimuli varying in representation type (symbolic vs. non-symbolic) and mode of stimuli delivery (visual vs. auditory). Time-frequency spectrograms were dimensionally reduced with principal component analysis and passed into a linear support vector machine classification algorithm to identify regions associated with number perception compared to inter-trial periods. Across representation formats, the highest classification accuracy was observed in the bilateral parietal lobes. Auditory (spoken and beeps) and visual (Arabic) number formats preferentially engaged the superior temporal cortices and the frontoparietal regions, respectively. The left parietal cortex was found to have the highest classification for number dots. Notably, the putamen exhibited robust classification accuracies in response to numerical stimuli. Analyses of spectral feature maps revealed that non-gamma frequency, below 30 Hz, had greater-than-chance classification value and could be potentially used to characterize format specific number representations. Taken together, our findings obtained from intracranial recordings provide further support and expand on the Triple Code Model for numerical cognition.

Broad Distributed Game Learning for intelligent classification in rolling bearing fault diagnosis

As a new Single Layer Feedforward Network (SLFN) architecture, Broad Learning System (BLS) has been widely used in the field of fault diagnosis because of its fast-training speed and high generalization capability. However, when features in different classes of signals are similar or weak, BLS generates a large number of redundant features that may be difficult to classify accurately. In view of this, a new Broad Distributed Game Learning (BDGL) method is proposed in this paper, which maps data into the game space by constructing two non-parallel game hyperplanes to achieve game and segmentation of different similar features, thereby making the data linearly differentiable in the game space. Meanwhile, a linear distribution constraint term is designed to reduce noise fitting and weak feature learning in training data learning by limiting the complexity of model parameters, thereby making the solution of the objective function simpler and faster. By comparing the Precision, Recall, F-score, Kappa and Accuracy of BDGL and the comparison methods on the two types of rolling bearing experimental data, the results show that BDGL has a high classification accuracy. In addition, the experimental results on small and noisy samples once again demonstrate the effectiveness of BDGL, which provides an efficient solution for rolling bearing fault diagnosis.

SBPA: Sybil-Based Backdoor Poisoning Attacks for Distributed Big Data in AIoT-Based Federated Learning System

Federated learning (FL) enables a great deal of distributed independent participants to collaborate in training without sharing data. Malicious adversary can poison the local model by backdoor poisoning attacks and utilize the characteristic that server cannot trace original data to make the poisoned model directly aggregated. Especially in the AIoT-FL network that generates large amounts of data in real-time, such an attack is more powerful. In this paper, we design a sybil-based backdoor poisoning attacks (SBPA) against the above vulnerability. Malicious participants inject backdoor triggers into distributed Big Data to covertly complete data poisoning. During subsequent iterative aggregation, the joint model activates the backdoor during testing to achieve misclassification for the backdoor images. Besides, malicious participants create some sybil nodes to join the aggregation by taking advantage of the vulnerability of system devices to be easily disconnected. They are committed to making the poisoned local models aggregated with higher probability. Their goal works on making the final global model misclassify backdoor images while keeping high classification accuracy on the other non-backdoor samples. We conduct extensive experiments on multiple datasets and exhibit more robust performance than the state-of-the-art in various metrics under both data distribution including i.i.d. and non-i.i.d. scenarios.

Comparison of radiated noise classification methods for underwater targets based on different enhanced images and convolutional neural networks

With the continuous development of economy and society, factors such as the variety of underwater targets and the high level of environmental noise have a great impact on the classification accuracy of underwater target radiation noise, and the traditional classification method based on signal features can no longer meet the requirements of underwater target identification. In this paper, we propose an underwater target radiation noise classification method based on enhanced image and convolutional neural network. First, the underwater target radiation noise signal is converted into enhanced image by various methods, then the converted image data set is used as the input of convolutional neural network for model training, and finally the great advantage of convolutional neural network in image classification is used to accurately classify underwater target radiation noise. In order to propose an optimal augmented image transformation method, this paper uses several augmented image transformation methods and compares the classification results. The experimental results show that the augmented image and convolutional neural network methods based on lagomorphs and corner fields have the highest classification accuracy and the best classification efficiency.

A Hardware-Efficient Novelty-Aware Spike Sorting Approach for Brain-Implantable Microsystems.

Unsupervised spike sorting, a vital processing step in real-time brain-implantable microsystems, is faced with the prominent challenge of managing nonstationarity in neural signals. In long-term recordings, spike waveforms gradually change and new source neurons are likely to become activated. Adaptive spike sorters combined with on-implant training units effectively process the nonstationary signals at the cost of high hardware resource utilization. On the other hand, static approaches, while being hardware-friendly, are subjected to decreased processing performance in such recordings where the neural signal characteristics gradually change. To strike a balance between the hardware cost and processing performance, this study proposes a hardware-efficient novelty-aware spike sorting approach that is capable of dealing with both variated spike waveforms and spike waveforms generated from new source neurons. Its improved hardware efficiency compared to adaptive ones and capability of dealing with nonstationary signals make it attractive for implantable applications. The proposed novelty-aware spike sorting especially would be a good fit for brain-computer interfaces where long-term, real-time interaction with the brain is required, and the available on-implant hardware resources are limited. Our unsupervised spike sorting benefits from a novelty detection process to deal with neural signal variations. It tracks the spike features so that in case of detecting an unexpected change (novelty detection) both on and off-implant parameters are updated to preserve the performance in new state. To make the proposed approach agile enough to be suitable for brain implants, the on-implant computations are reduced while the computational burden is realized off-implant. The performance of our proposed approach is evaluated using both synthetic and real datasets. The results demonstrate that, in the mean, it is capable of detecting 94.31% of novel spikes (wave-drifted or emerged spikes) with a classification accuracy (CA) of 96.31%. Moreover, an FPGA prototype of the on-implant circuit is implemented and tested. It is shown that in comparison to the OSORT algorithm, a pivotal spike sorting method, our spike sorting provides a higher CA at significantly lower hardware resources. The proposed circuit is also implemented in a 180-nm standard CMOS process, achieving a power consumption of 1.78[Formula: see text][Formula: see text] per channel and a chip area of 0.07[Formula: see text]mm2 per channel.

Exploring the Efficacy of Various CNN Architectures in Diagnosing Oral Cancer from Squamous Cell Carcinoma

Oral cancer can result from mutations in cells located in the lips or mouth. Diagnosing oral cavity squamous cell carcinoma (OCSCC) is particularly challenging, often occurring at advanced stages. To address this, computer-aided diagnosis methods are increasingly being used. In this work, a deep learning-based approach utilizing models such as VGG16, ResNet50, LeNet-5, MobileNetV2, and Inception V3 is presented. NEOR and OCSCC datasets were used for feature extraction, with virtual slide images divided into tiles and classified as normal or squamous cell cancer. Performance metrics like accuracy, F1-score, AUC, precision, and recall were analyzed to determine the prerequisites for optimal CNN performance. The proposed CNN approaches were effective for classifying OCSCC and oral dysplasia, with the highest accuracy of 95.41% achieved using MobileNetV2. Key findingsDeep learning models, particularly MobileNetV2, achieved high classification accuracy (95.41%) for OCSCC.CNN-based methods show promise for early-stage OCSCC and oral dysplasia diagnosis. Performance parameters like precision, recall, and F1-score help optimize CNN model selection for this task.

Potential impact of organophosphate esters on thyroid eye disease based on machine learning and molecular docking

Organophosphate esters (OPEs) are widely used as flame retardants and plasticizers in daily commodities and building materials. Some OPEs, acting as agonists of the thyroid-stimulating hormone receptor (TSHR), may contribute to the development of thyroid eye disease (TED). This study analyzes the serum and urine of patients and control groups, using machine learning and molecular docking to investigate the potential impact of OPEs on TED. Results indicate significantly higher concentrations of OPEs and di-OPEs of TED patients compared to controls (Mann-Whitney U test, p < 0.05). Aryl OPEs exhibit the strongest binding affinity with TSHR. We developed a predictive model for OPE-TSHR affinity to explore the impact of OPE structural features on TSHR activity and effectively capture the complex relationships between changes in OPE side chains and their effects on TSHR. Predictions from the USEPA's database indicate that 28 % of 1011 OPEs have a tendency to bind with TSHR. Furthermore, a high-accuracy classification model successfully identified key substructures associated with high affinity for TSHR. This study not only enhances our understanding of the complex relationship between the structural diversity of OPEs and their thyroid impact but also offers molecular design insights to prevent releasing OPEs with high thyroid harm potential into the environment.

Feature Selection and Classification for Anomaly Detection in IoT Networks using Multi-layer Perceptron - Chaotic Ant Lion Optimization (MLP-CALO) Algorithm

As the number of devices in Internet of Things (IoT) increases exponentially, the risks of unknown vulnerabilities and threats also increases resulting in degradation of performance. Traditional anomaly detection systems are ineffective within IoT ecosystems, since the range of possible normal behaviours of devices is significantly larger and more dynamic. The main objective of this work is to develop a machine learning (ML) algorithm to classify the selected features of IoT traffic and detect the malicious behaviour of users and unauthorised devices dynamically. In this paper feature Selection and classification for anomaly detection in IoT using Multi Layer Perception (MLP) and Chaotic Ant Lion Optimization (CALO) algorithm is proposed. In this algorithm, Filtering method using Fisher’ score and correlation coefficient is applied to select the candidate feature set. Then hybrid MLP-CALO algorithm is proposed to classify the selected features and detects the anomalies from the IoT traffic. The objective function of the CALO algorithm minimizes the average MSE of MLP output. Experimental results have shown that MLP-CALO has higher classification accuracy and lesser computation cost when compared to traditional ALO and ANN algorithms

Multichannel EMG-based gesture recognition utilizing advanced machine learning techniques: A random forest classifier for high-precision signal classification

This research examines how advanced machine learning algorithms can be used to classify multichannel electromyographic (EMG) signals with a high level of accuracy to assist in recognizing hand gestures. The goal is to create a robust and scalable system for gesture-based virtual control using EMG signals with potential applications in assistive technologies, rehabilitation, and human-computer interaction. Data were gathered using a MYO Thalmic bracelet containing eight EMG sensors on thirty-six subjects, and a Random Forest classifier was trained to identify seven distinct types of hand gestures (rest, fist clench, wrist flexion/extension, and radial/ulnar deviations). The machine learning pipeline included extensive preprocessing (i.e., EMG signal normalization and signal feature extraction; root mean square, waveform length, and zero crossing rate) and several hyperparameter tuning procedures to improve model performance. The Random Forest model (100 decision trees) achieved an overall classification accuracy of 98.68%, with a range of accuracies for each class (e.g., 95.2% wrist flexion and 91.8% ulnar deviation) when evaluated using cross-validation (i.e., average F1-score = 0.92, precision = 0.94, recall = .91). Overall, the study provides strong evidence for the effectiveness of ensemble learning methods at analyzing complex, multidimensional EMG signals. The high classification accuracy reported, in particular, demonstrates that the system could function for real-time recognition of hand gestures in a virtual environment. Ultimately, the initial work sets the stage for future exploration of a model that may be integrated with actuation models to control prosthetic limbs, virtual actors/avatars, and robotic devices. By demonstrating a scalable and efficient method of gesture recognition using EMG signals, these early findings enable future pathways and possibilities to design innovative, assistive solutions for digital systems that increase accessibility and interaction for users who are motor impaired or have a limited range of motion.

A Study on Filter-Based Adversarial Image Classification Models

In view of the fact that adversarial examples can lead to high-confidence erroneous outputs of deep neural networks, this study aims to improve the safety of deep neural networks by distinguishing adversarial examples. A classification model based on filter residual network structure is used to accurately classify adversarial examples. The filter-based classification model includes residual network feature extraction and classification modules, which are iteratively optimized by an adversarial training strategy. Three mainstream adversarial attack methods are improved, and adversarial samples are generated on the Mini-ImageNet dataset. Subsequently, these samples are used to attack the EfficientNet and the filter-based classification model respectively, and the attack effects are compared. Experimental results show that the filter-based classification model has high classification accuracy when dealing with Mini-ImageNet adversarial examples. Adversarial training can effectively enhance the robustness of deep neural network models.

Non-Destructive Monitoring of External Quality of Date Palm Fruit (Phoenix dactylifera L.) During Frozen Storage Using Digital Camera and Flatbed Scanner.

The emergence of new technologies focusing on "computer vision" has contributed significantly to the assessment of fruit quality. In this study, an innovative approach based on image analysis was used to assess the external quality of fresh and frozen 'Mejhoul' and 'Boufeggous' date palm cultivars stored for 6 months at -10 °C and -18 °C. Their quality was evaluated, in a non-destructive manner, based on texture features extracted from images acquired using a digital camera and flatbed scanner. The whole process of image processing was carried out using MATLAB R2024a and Q-MAZDA 23.10 software. Then, extracted features were used as inputs for pre-established algorithms-groups within WEKA 3.9 software to classify frozen date fruit samples after 0, 2, 4, and 6 months of storage. Among 599 features, only 5 to 36 attributes were selected as powerful predictors to build desired classification models based on the "Functions-Logistic" classifier. The general architecture exhibited clear differences in classification accuracy depending mainly on the frozen storage period and imaging device. Accordingly, confusion matrices showed high classification accuracy (CA), which could reach 0.84 at M0 for both cultivars at the two frozen storage temperatures. This CA indicated a remarkable decrease at M2 and M4 before re-increasing by M6, confirming slight changes in external quality before the end of storage. Moreover, the developed models on the basis of flatbed scanner use allowed us to obtain a high correctness rate that could attain 97.7% in comparison to the digital camera, which did not exceed 85.5%. In perspectives, physicochemical attributes can be added to developed models to establish correlation with image features and predict the behavior of date fruit under storage.