Convolutional Neural Network Transfer Learning Research Articles

Research on sound quality of roller chain transmission system based on multi-source transfer learning

To establish the sound quality evaluation model of roller chain transmission system, we collect the running noise under different working conditions. After the noise samples are preprocessed, a group of experienced testers are organized to evaluate them subjectively. Mel frequency cepstral coefficient (MFCC) of each noise sample is calculated, and the MFCC feature map is used as an objective evaluation. Combining with the subjective and objective evaluation results of the roller chain system noise, we can get the original dataset of its sound quality research. However, the number of high-quality noise samples is relatively small. Based on the sound quality research of various chain transmission systems, a novel method called multi-source transfer learning convolutional neural network (MSTL-CNN) is proposed. By transferring knowledge from multiple source tasks to target task, the difficulty of small sample sound quality prediction is solved. Compared with the problem that single source task transfer learning has too much error on some samples, MSTL-CNN can give full play to the advantages of all transfer learning models. The results also show that the MSTL-CNN proposed in this paper is significantly better than the traditional sound quality evaluation methods.

Centrifugal Pump Fault Detection with Convolutional Neural Network Transfer Learning.

The centrifugal pump is the workhorse of many industrial and domestic applications, such as water supply, wastewater treatment and heating. While modern pumps are reliable, their unexpected failures may jeopardise safety or lead to significant financial losses. Consequently, there is a strong demand for early fault diagnosis, detection and predictive monitoring systems. Most prior work on machine learning-based centrifugal pump fault detection is based on either synthetic data, simulations or data from test rigs in controlled laboratory conditions. In this research, we attempted to detect centrifugal pump faults using data collected from real operational pumps deployed in various places in collaboration with a specialist pump engineering company. The detection was done by the binary classification of visual features of DQ/Concordia patterns with residual networks. Besides using a real dataset, this study employed transfer learning from the image detection domain to systematically solve a real-life problem in the engineering domain. By feeding DQ image data into a popular and high-performance residual network (e.g., ResNet-34), the proposed approach achieved up to 85.51% classification accuracy.

An investigation into real-time surface crack classification and measurement for structural health monitoring using transfer learning convolutional neural networks and Otsu method

This study introduces a pioneering system for real-time classification and measurement of concrete surface cracks, a crucial aspect of Structural Health Monitoring (SHM). We harness the power of transfer learning (TL) in Convolutional Neural Networks (CNNs), including renowned models such as MobileNetV2, EfficientNetV2, InceptionV3, and ResNet50. Notably, our model excels, particularly with TL MobileNetV2, achieving remarkable results – a 99.87% accuracy rate, 99.74% recall, 100% precision, and an impressive 99.87% F1-score. Incorporating the Otsu method for image segmentation, our system accurately assesses individual crack sizes. To refine measurements, Euclidean distance calculations and a 'pixel per inch' technique, accounting for video resolution, ensure millimeter-level width estimations. Precision is validated through manual experiments using a vernier caliper, specifically the Mitutoyo Absolute Digital Caliper. This tool ensures high accuracy, with an error margin of ±0.2 mm to ±0.3 mm, making it efficient for detailed measurements. Despite these promising outcomes, it is crucial to acknowledge inherent limitations. These include dependence on image quality, challenges in generalization, sensitivity to training data, assumption of linear crack width calculation, resolution dependency, and other factors. These limitations underscore the need for further refinement in our proposed classification model and measurement technique. This research represents a significant advancement in the SHM field, catering to early detection and timely maintenance requirements essential for infrastructure safety and longevity. As the field increasingly prioritizes rapid detection, our model presents a versatile solution that enhances the potential of Structural Health Monitoring.

Prediction of ash content in coal slime flotation based on CNN-BP method with residual estimation

ABSTRACT Ash content is a key metric for assessing coal slime flotation performance. Traditional methods of sampling, combustion, and weighing lead to measurement delays, thus image-based real-time ash prediction during flotation has become meaningful. However, the complex characteristics of the multiphase system in flotation pulp images make the extraction of an effective dataset for ash recognition challenging. This paper introduces an image-based method for predicting ash content in solid-liquid pulp mixtures using Convolutional Neural Networks (CNNs). A CNN transfer learning method based on ImageNet is proposed, combined with a Backpropagation (BP) residual estimation module with specified feature filters, reducing the sample size requirement. Four classic CNN networks were utilized. Results indicate that introducing the BP residual estimation module improves the CNN network’s prediction accuracy in small samples, with total residuals consistent with the white noise distribution. Notably, the 5% accuracy (absolute error less than 5%) of VGG19 without the CNN classic residual structure improved significantly, reflecting the superior residual estimation of BP and the robust performance of the CNN-BP in predicting pulp ash content. This research offers a new option for transitioning previous dataset collection methods for ash detection in the coal slime flotation process.

Tropical cyclone intensity estimation through convolutional neural network transfer learning using two geostationary satellite datasets

Accurate prediction and monitoring of tropical cyclone (TC) intensity are crucial for saving lives, mitigating damages, and improving disaster response measures. In this study, we used a convolutional neural network (CNN) model to estimate TC intensity in the western North Pacific using Geo-KOMPSAT-2A (GK2A) satellite data. Given that the GK2A data cover only the period since 2019, we applied transfer learning to the model using information learned from previous Communication, Ocean, and Meteorological Satellite (COMS) data, which cover a considerably longer period (2011–2019). Transfer learning is a powerful technique that can improve the performance of a model even if the target task is based on a small amount of data. Experiments with various transfer learning methods using the GK2A and COMS data showed that the frozen–fine-tuning method had the best performance due to the high similarity between the two datasets. The test results for 2021 showed that employing transfer learning led to a 20% reduction in the root mean square error (RMSE) compared to models using only GK2A data. For the operational model, which additionally used TC images and intensities from 6 h earlier, transfer learning reduced the RMSE by 5.5%. These results suggest that transfer learning may represent a new breakthrough in geostationary satellite image–based TC intensity estimation, for which continuous long-term data are not always available.

EMG-based Multi-User Hand Gesture Classification via Unsupervised Transfer Learning Using Unknown Calibration Gestures.

The poor generalization performance and heavy training burden of the gesture classification model contribute as two main barriers that hinder the commercialization of sEMG-based human-machine interaction (HMI) systems. To overcome these challenges, eight unsupervised transfer learning (TL) algorithms developed on the basis of convolutional neural networks (CNNs) were explored and compared on a dataset consisting of 10 gestures from 35 subjects. The highest classification accuracy obtained by CORrelation Alignment (CORAL) reaches more than 90%, which is 10% higher than the methods without using TL. In addition, the proposed model outperforms 4 common traditional classifiers (KNN, LDA, SVM, and Random Forest) using the minimal calibration data (two repeated trials for each gesture). The results also demonstrate the model has a great transfer robustness/flexibility for cross-gesture and cross-day scenarios, with an accuracy of 87.94% achieved using calibration gestures that are different with model training, and an accuracy of 84.26% achieved using calibration data collected on a different day, respectively. As the outcomes confirm, the proposed CNN TL method provides a practical solution for freeing new users from the complicated acquisition paradigm in the calibration process before using sEMG-based HMI systems.

Acoustic emission reflection signal classification of PVDF-type AE sensor using convolutional neural network-transfer learning

Acoustic emission reflection signal classification of PVDF-type AE sensor using convolutional neural network-transfer learning

Deep transfer learning CNN based for classification quality of organic vegetables

This study introduces a system based on a Convolutional Neural Network (CNN) with deep transfer learning for classifying organic vegetables. It aims to evaluate their quality through artificial intelligence. The approach involves three key steps: collecting data, preparing data, and creating data models. Initially, the data collection phase involves gathering images of organic vegetables from packing facilities, organizing these images into training, testing, and validation datasets. In the preparation phase, image processing techniques are applied to adjust the images for training and testing, resizing each to 224 x 224 pixels. The modeling phase involves using these prepared datasets, which include 3,239 images of two types of organic vegetables, to train the model. The study tests the model's effectiveness using three CNN architectures: Inception V3, VGG16, and ResNet50. It finds that the Inception V3 model achieves the highest accuracy at 85%, VGG16 follows with 82% accuracy, and ResNet50 has the lowest accuracy at 50%. The results suggest that Inception V3 is the most effective at accurately classifying organic vegetables, while VGG16 shows some limitations in certain categories, and ResNet50 is the least effective.

Non-invasive detection of anemia using lip mucosa images transfer learning convolutional neural networks.

Anemia is defined as a drop in the number of erythrocytes or hemoglobin concentration below normal levels in healthy people. The increase in paleness of the skin might vary based on the color of the skin, although there is currently no quantifiable measurement. The pallor of the skin is best visible in locations where the cuticle is thin, such as the interior of the mouth, lips, or conjunctiva. This work focuses on anemia-related pallors and their relationship to blood count values and artificial intelligence. In this study, a deep learning approach using transfer learning and Convolutional Neural Networks (CNN) was implemented in which VGG16, Xception, MobileNet, and ResNet50 architectures, were pre-trained to predict anemia using lip mucous images. A total of 138 volunteers (100 women and 38 men) participated in the work to develop the dataset that contains two image classes: healthy and anemic. Image processing was first performed on a single frame with only the mouth area visible, data argumentation was preformed, and then CNN models were applied to classify the dataset lip images. Statistical metrics were employed to discriminate the performance of the models in terms of Accuracy, Precision, Recal, and F1 Score. Among the CNN algorithms used, Xception was found to categorize the lip images with 99.28% accuracy, providing the best results. The other CNN architectures had accuracies of 96.38% for MobileNet, 95.65% for ResNet %, and 92.39% for VGG16. Our findings show that anemia may be diagnosed using deep learning approaches from a single lip image. This data set will be enhanced in the future to allow for real-time classification.

A Novel Shipyard Production State Monitoring Method Based on Satellite Remote Sensing Images

Monitoring the shipyard production state is of great significance to shipbuilding industry development and coastal resource utilization. In this article, it is the first time that satellite remote sensing (RS) data is utilized to monitor the shipyard production state dynamically and efficiently, which can make up for the traditional production state data collection mode. According to the imaging characteristics of optical remote sensing images in shipyards with a different production state, the characteristics are analyzed to establish reliable production state evidence. Firstly, in order to obtain the characteristics of the production state of optical remote sensing data, the high-level semantic information in the shipyard is extracted by transfer learning convolutional neural networks (CNNs). Secondly, in the evidence fusion, for the conflict evidence from the core sites of the shipyard, an improved DS evidence fusion method is proposed, which constructs the correlation metric to measure the degree of conflict in evidence and designs the similarity metric to measure the credibility of evidence. Thirdly, the weight of all the evidence is calculated according to the similarity metric to correct the conflict evidence. The introduction of the iterative idea is motivated by the fact that the fusion result aligns more closely with the desired result, the iterative idea is introduced to correct the fusion result. This method can effectively solve the conflict of evidence and effectively improve the monitoring accuracy of the shipyard production state. In the experiments, the Yangtze River Delta and the Bohai Rim are selected to verify that the proposed method can accurately recognize the shipyard production state, which reveals the potential of satellite RS images in shipyard production state monitoring, and also provides a new research thought perspective for other industrial production state monitoring.

Pseudo-labeling of transfer learning convolutional neural network data for human facial emotion recognition

The relevance of solving the problem of facial emotion recognition on human images in the creation of modern intelligent systems of computer vision and human-machine interaction, online learning and emotional marketing, health care and forensics, machine graphics and game intelligence is shown. Successful examples of technological solutions to the problem of facial emotion recognition using transfer learning of deep convolutional neural networks are shown. But the use of such popular datasets as DISFA, CelebA, AffectNet, for deep learning of convolutional neuralnetworks does not give good results in terms of the accuracy of emotion recognition, because almost all training sets have fundamental flaws related to errors in their creation, such as the lack of data of a certain class, imbalance of classes, subjectivity and ambiguity of labeling, insufficient amount of data for deep learning, etc. It is proposed to overcome the noted shortcomings of popular datasets for emotion recognition by adding to the training sample additional pseudo-labeled images with human emotions, on which recognition occurs with high accuracy. The aim of the research is to increase the accuracy of facial emotion recognitionon the image of a human by developing a pseudo-labeling method for transfer learning of a deep neural network. To achieve the aim, the following tasks were solved: a convolutional neural network model, previously trained on the ImageNet set using the transfer learning method, was adjusted on the RAF-DB data set to solve emotion recognition tasks; a pseudo-labeling method of the RAF−DB set data was developed for semi-supervised learning of a convolutional neural network model for the task of facial emotion recognition; the accuracy of facial emotion recognition was analyzed based on the developed convolutional neural network model and the method of pseudo-labeling of RAF-DB set data for its correction. It is shown that the use of the developed method of pseudo-labeling data and transfer learning of the MobileNet V1 convolutional neural network model allowed to increase the accuracy of facial emotion recognitionon the images of the RAF-DB dataset by 2 percent (from 76 to 78%) according to the F1 estimate. Atthe same time, taking into account the significant imbalance of the classes, for the 7 main emotions in the trainingset, we have a significant increase in the accuracy of recognizing a few representatives of such emotions as surprise (from 71 to 77%), fearful(from 64 to 69%), sad (from 72 to 76%), angrywith (from 64 to 74%), neutral(from 66 to 71%). The accuracy of recognizing the emotion of happy, which is the most common, decreased (from 91 to 86 %) Thus, it can be concluded that the use of the developed pseudo-labeling method gives good results in overcoming such shortcomings of datasets for deep learning of convolutional neural networks such as lack of data of a certain type, imbalance of classes, insufficient amount of data for deep learning, etc.

Classifying enset based on their disease tolerance using deep learning

Even though agriculture remains the main source of Ethiopia’s population economy, not identifying enset based on their disease tolerance level is an obstacle for the production of enset.This paper’s main objective is to automatically identify the disease resistance levels of enset plants through digital image. The researcher followed the design science research method to achieve the objective listed above. Besides, the researcher has attempted to get valuable information about the type and the nature of these classes from the domain expert through interviews, document analysis, and observation from the fields. The total number of images used for experimentation purposes was 3000. The Contaharmonic filtering technique was implemented to remove noise due to its highest entropy recorded. A deep learning-based approach with training from scratch and transfer learning convolutional neural network methods were applied. Based on this, the researcher made experimentation for transfer learning by using two different pre-trained models, namely VGG-19 and VGG-16. Finally, the developed classifier model’s performance was assessed using accuracy, precision, recall, and the F1 score. According to the interpretation of the results, the proposed model’s training from scratch method achieves 92.6%. On the other way, the accuracy obtained with the transfer learning method, VGG-16 achieves 98.5%, and VGG-19 achieves 93.9%. Hence, transfer learning, specifically the VGG-16 model revealed an effective and robust performance for classifying enset based on their disease tolerance level based on the researcher’s number of images.

Remote Sensing Image Scene Classification in Hybrid Classical-Quantum Transferring CNN with Small Samples.

The scope of this research lies in the combination of pre-trained Convolutional Neural Networks (CNNs) and Quantum Convolutional Neural Networks (QCNN) in application to Remote Sensing Image Scene Classification(RSISC). Deep learning (RL) is improving by leaps and bounds pretrained CNNs in Remote Sensing Image (RSI) analysis, and pre-trained CNNs have shown remarkable performance in remote sensing image scene classification (RSISC). Nonetheless, CNNs training require massive, annotated data as samples. When labeled samples are not sufficient, the most common solution is using pre-trained CNNs with a great deal of natural image datasets (e.g., ImageNet). However, these pre-trained CNNs require a large quantity of labelled data for training, which is often not feasible in RSISC, especially when the target RSIs have different imaging mechanisms from RGB natural images. In this paper, we proposed an improved hybrid classical-quantum transfer learning CNNs composed of classical and quantum elements to classify open-source RSI dataset. The classical part of the model is made up of a ResNet network which extracts useful features from RSI datasets. To further refine the network performance, a tensor quantum circuit is subsequently employed by tuning parameters on near-term quantum processors. We tested our models on the open-source RSI dataset. In our comparative study, we have concluded that the hybrid classical-quantum transferring CNN has achieved better performance than other pre-trained CNNs based RSISC methods with small training samples. Moreover, it has been proven that the proposed algorithm improves the classification accuracy while greatly decreasing the amount of model parameters and the sum of training data.

Identifikasi Tumor Otak Citra MRI dengan Convolutional Neural Network

The science of artificial intelligence and computer vision is beneficial in facilitating the detection of diseases in the medical field. Computer-based disease detection can save time. However, identifying and detecting tumors on MRI images require seriousness and is time-consuming. Due to the diversity of structures in size, shape, and intensity of the image, accuracy is needed in identifying the original organ structure and the diseased one. Previous studies have proposed a method for identifying brain tumors to produce the correct precision. In previous studies, neural network-based methods have good accuracy. We present five Convolutional Neural Network (CNN) architectures for identifying brain tumors (glioma, meningioma, no tumor, and pituitary) on MRI images. This study aims to develop an optimal CNN architecture for identifying tumors. We use the dataset from Kaggle with a total training data of 5712 and testing of 1311. Of the five proposed CNN architectures, architecture c has the highest accuracy of 82.2% with an unlimited number of parameters of 29605060. A good CNN architecture has many convolution layers. We also compare the proposed architecture with CNN transfer learning (Inception, ResNet-50, and VGG16), and with CNN transfer learning architecture, the accuracy is higher than our proposed architecture.

Deep Learning-Based Wildfire Image Detection and Classification Systems for Controlling Biomass

Forests are essential natural resources that directly impact the ecosystem. However, the rising frequency of forest fires due to natural and artificial climate change has become a critical issue. A revolutionary municipal application proposes deploying an artificial intelligence-based forest fire warning system to prevent major disasters. This work aims to present an overview of vision-based methods for detecting and categorizing forest fires. The study employs a forest fire detection dataset to address the classification difficulty of discriminating between photos with and without fire. This method is based on convolutional neural network transfer learning with Inception-v3. Thus, automatic identification of current forest fires (including burning biomass) is a critical field of research for reducing negative repercussions. Early fire detection can also assist decision-makers in developing mitigation and extinguishment strategies. Radial basis function Networks (RBFNs) with rapid and accurate image super resolution (RAISR) is a deep learning framework trained on an input dataset to detect active fires and burning biomass. The proposed RBFN-RAISR model’s performance in recognizing fires and nonfires was compared to earlier CNN models using several performance criteria. The water wave optimization technique is used for image feature selection, noise and blurring reduction, image improvement and restoration, and image enhancement and restoration. When classifying fire and no-fire photos, the proposed RBFN-RAISR fire detection approach achieves 97.55% accuracy, 93.33% F-Score, 96.44% recall, 94.19% precision, and an error rate of 24.89. Given the one-of-a-kind forest fire detection dataset, the suggested method achieves promising results for the forest fire categorization problem.

Finger pinching and imagination classification: A fusion of CNN architectures for IoMT-enabled BCI applications

A Brain–Computer Interface (BCI), integrated with the Internet of Medical Things (IoMT) and based on electroencephalogram (EEG) technology, allows users to control external devices by decoding brainwave patterns. Advanced deep learning-based BCIs, especially those utilizing sensorimotor rhythms (SMRs), have emerged as direct brain–device communication facilitators. SMRs involve users imagining limb motions to induce specific brain activity changes in the motor cortex. Despite progress, some users struggle with BCIs due to weak signals, individual variability, and limited task applicability. This study introduces an unsupervised EEG preprocessing pipeline for SMR-based BCIs. It evaluates an EEG dataset recorded during finger movements, employing two cleaning methods: an investigator-dependent pipeline and our proposed unsupervised method. Two distinct feature datasets are generated: one from cleaned EEG data processed into spectrogram images using supervised preprocessing, and another from data cleaned using our proposed unsupervised pipeline. The study extensively assesses five transfer learning convolutional neural network (TL-CNN) models for distinguishing Motor Imagery (MI) from finger movements (Mex) using the generated datasets. A novel probability fusion technique is developed to enhance TL-CNN classification in Mex versus MI finger-pinching actions. Comparative results show that the fusion-based method outperforms other state-of-the-art methods when applied to unsupervised EEG data. Specifically, our proposed approach achieves 97.9% accuracy, 93.4% precision, 95% recall, and an F1-score of 93.2%, demonstrating significant progress in distinguishing MI and Mex activities through the use of our unsupervised pre-processing pipeline and fusion-based CNN method. Our findings demonstrate the potential of our approach to serve as a benchmark for the global interdisciplinary research community and enable the development of future more effective and user-friendly real-time BCI systems.

Classical-to-quantum convolutional neural network transfer learning

Machine learning using quantum convolutional neural networks (QCNNs) has demonstrated success in both quantum and classical data classification. In previous studies, QCNNs attained a higher classification accuracy than their classical counterparts under the same training conditions in the few-parameter regime. However, the general performance of large-scale quantum models is difficult to examine because of the limited size of quantum circuits, which can be reliably implemented in the near future. We propose transfer learning as an effective strategy for utilizing small QCNNs in the noisy intermediate-scale quantum era to the full extent. In the classical-to-quantum transfer learning framework, a QCNN can solve complex classification problems without requiring a large-scale quantum circuit by utilizing a pre-trained classical convolutional neural network (CNN). We perform numerical simulations of QCNN models with various sets of quantum convolution and pooling operations for MNIST data classification under transfer learning, in which a classical CNN is trained with Fashion-MNIST data. The results show that transfer learning from classical to quantum CNN performs considerably better than purely classical transfer learning models under similar training conditions.

Classification of Appearance Quality of Red Grape Based on Transfer Learning of Convolution Neural Network

Grapes are a globally popular fruit, with grape cultivation worldwide being second only to citrus. This article focuses on the low efficiency and accuracy of traditional manual grading of red grape external appearance and proposes a small-sample red grape external appearance grading model based on transfer learning with convolutional neural networks (CNNs). In the naturally growing vineyards, 195,120,135 samples of Grade I, Grade II, and Grade III red grapes were collected using a Canon EOS 550D camera, and a data set of 1800 samples was obtained using data enhancement technology. Then, the CNN transfer learning method was used to transfer the pre-trained AlexNet, VGG16, GoogleNet, InceptionV3, and ResNet50 network models on the ImageNet image dataset to the red grape image grading task. By comparing the classification performance of the CNN models of these five different network depths with fine-tuning, ResNet50 with a learning rate of 0.001 and a loop number of 10 was determined to be the best feature extractor for red grape images. Moreover, given the small number of red grape image samples in this study, different convolutional layer features output by the ResNet50 feature extractor were analyzed layer by layer to determine the effect of deep features extracted by each convolutional layer on Support Vector Machine (SVM) classification performance. This analysis helped to obtain a ResNet50 + SVM red grape external appearance grading model based on the optimal ResNet50 feature extraction strategy. Experimental data showed that the classification model constructed using the feature parameters extracted from the 10th node of the ResNet50 network achieved an accuracy rate of 95.08% for red grape grading. These research results provide a reference for the online grading of red grape clusters based on external appearance quality and have certain guiding significance for the quality and efficiency of grape industry circulation and production.

Comparison of CNN-based deep learning architectures for rice diseases classification

Although convolutional neural network (CNN) paradigms have expanded to transfer learning and ensemble models from original individual CNN architectures, few studies have focused on the performance comparison of the applicability of these techniques in detecting and localizing rice diseases. Moreover, most CNN-based rice disease detection studies only considered a small number of diseases in their experiments. Both these shortcomings were addressed in this study. In this study, a rice disease classification comparison of six CNN-based deep-learning architectures (DenseNet121, Inceptionv3, MobileNetV2, resNext101, Resnet152V, and Seresnext101) was conducted using a database of nine of the most epidemic rice diseases in Bangladesh. In addition, we applied a transfer learning approach to DenseNet121, MobileNetV2, Resnet152V, Seresnext101, and an ensemble model called DEX (Densenet121, EfficientNetB7, and Xception) to compare the six individual CNN networks, transfer learning, and ensemble techniques. The results suggest that the ensemble framework provides the best accuracy of 98%, and transfer learning can increase the accuracy by 17% from the results obtained by Seresnext101 in detecting and localizing rice leaf diseases. The high accuracy in detecting and categorisation rice leaf diseases using CNN suggests that the deep CNN model is promising in the plant disease detection domain and can significantly impact the detection of diseases in real-time agricultural systems. This research is significant for farmers in rice-growing countries, as like many other plant diseases, rice diseases require timely and early identification of infected diseases and this research develops a rice leaf detection system based on CNN that is expected to help farmers to make fast decisions to protect their agricultural yields and quality.

AI-enabled case detection model for infectious disease outbreaks in resource-limited settings

IntroductionThe utility of non-contact technologies for screening infectious diseases such as COVID-19 can be enhanced by improving the underlying Artificial Intelligence (AI) models and integrating them into data visualization frameworks. AI models that are a fusion of different Machine Learning (ML) models where one has leveraged the different positive attributes of these models have the potential to perform better in detecting infectious diseases such as COVID-19. Furthermore, integrating other patient data such as clinical, socio-demographic, economic and environmental variables with the image data (e.g., chest X-rays) can enhance the detection capacity of these models.MethodsIn this study, we explore the use of chest X-ray data in training an optimized hybrid AI model based on a real-world dataset with limited sample size to screen patients with COVID-19. We develop a hybrid Convolutional Neural Network (CNN) and Random Forest (RF) model based on image features extracted through a CNN and EfficientNet B0 Transfer Learning Model and applied to an RF classifier. Our approach includes an intermediate step of using the RF's wrapper function, the Boruta Algorithm, to select important variable features and further reduce the number of features prior to using the RF model.Results and discussionThe new model obtained an accuracy and recall of 96% for both and outperformed the base CNN model and four other experimental models that combined transfer learning and alternative options for dimensionality reduction. The performance of the model fares closely to relatively similar models previously developed, which were trained on large datasets drawn from different country contexts. The performance of the model is very close to that of the “gold standard” PCR tests, which demonstrates the potential for use of this approach to efficiently scale-up surveillance and screening capacities in resource limited settings.