Huge Amount Of Training Data Research Articles

Advanced Denoising and Meta-Learning Techniques for Enhancing Smart Health Monitoring Using Wearable Sensors

This study introduces a novel meta-learning method to enhance diabetes detection using wearable sensor systems in smart health applications. Wearable sensor technology often needs to operate accurately across a wide range of users, each characterized by unique physiological and behavioral patterns. However, the specific data for a particular application or user group might be scarce. Moreover, collecting extensive training data from wearable sensor experiments is challenging, time-consuming, and expensive. In these cases, meta-learning can be particularly useful. This model can quickly adapt to the nuances of new users or specific applications with minimal data. Therefore, to solve the need for a huge amount of training data and to enable the application of artificial intelligence (AI) in data-scarce scenarios, a meta-learning method is proposed. This meta-learning model has been implemented to forecast diabetes, resolve cross-talk issues, and accurately detect R peaks from overlapping electrocardiogram (ECG) signals affected by movement artifacts, poor electrode contact, electrical interference, or muscle activity. Motion artifacts from body movements, external conditions such as temperature, humidity, and electromagnetic interference, and the inherent quality and calibration of the sensor can all contribute to noise. Contact quality between the sensor and the skin, signal processing errors, power supply variations, user-generated interference from activities like talking or exercising, and the materials used in the wearable device also play significant roles in the overall noise in wearable sensor data and can significantly distort the true signal, leading to erroneous interpretations and potential diagnostic errors. Furthermore, discrete wavelet transform (DWT) was also implemented to improve the quality of the data and enhance the performance of the proposed model. The demonstrated results confirmed that with only a limited amount of target data, the proposed meta-learning and DWT denoising method can adapt more quickly and improve the detection of diabetes compared to the traditional method. Therefore, the proposed system is cost-effective, flexible, faster, and adaptable, reduces the need for training data, and can enhance the accuracy of chronic disease detection such as diabetes for smart health systems.

Few shot learning for avocado maturity determination from microwave images

Artificial intelligence (AI) has played a critical role in the enhancement and automation of global food production and delivery. The assessment of ripeness through microwave images and AI presents a unique opportunity to enhance the current process and minimize food wastage. Particularly in avocado maturity prediction, microwave imaging provides a non-invasive solution to see through the shell and predict the maturity intelligently. However, conventional deep learning models are generally data-hungry, requiring huge amounts of training data to maintain high performance. In this paper, we propose a few-shot learning (FSL) model to address the need for a large training dataset in maturity estimation. We classify avocado microwave images for maturity assessment through a prototypical FSL model. The FSL model achieves an accuracy range of 80%–96 % across the different experimental groups and all outperforms the conventional deep learning in scenarios where limited data is available for training. This experiment demonstrates the feasibility and accuracy of utilizing microwave scanning and FSL to determine avocado maturity.

Feed-forward networks using logistic regression and support vector machine for whole-slide breast cancer histopathology image classification

The performance of an image classification depends on the efficiency of the feature learning process. This process is a challenging task that traditionally requires prior knowledge from domain experts. Recently, representation learning was introduced to extract features directly from the raw images without any prior knowledge. Deep learning using a Convolutional Neural Network (CNN) has gained massive attention for performing image classification, as it achieves remarkable accuracy that sometimes exceeds human performance. But this type of network learns features by using a back-propagation approach. This approach requires a huge amount of training data and suffers from the vanishing gradient problem that deteriorates the feature learning. The forward-propagation approach uses predefined filters or filters learned outside the model and applied in a feed-forward manner. This approach is proven to achieve good results with small size labeled datasets. In this work, we investigate the suitability of using two feed-forward methods such as Convolutional Logistic Regression Network (CLR), and Convolutional Support Vector Machine Network for Histopathology Images (CSVM-H). The experiments we have conducted on two small breast cancer datasets (Sultan Qaboos University Hospital (SQUH) and BreaKHis dataset) demonstrate the advantage of using feed-forward approaches over the traditional back-propagation ones. On those datasets, the proposed models CLR and CSVM-H were faster to train and achieved better classification performance than the traditional back-propagation methods (VggNet-16 and ResNet-50) on the SQUH dataset. Importantly, our proposed approach CLR and CSVM-H efficiently learn representations from small amounts of breast cancer whole-slide images and achieve an AUC of 0.83 and 0.84, respectively, on the SQUH dataset. Moreover, the proposed models reduce memory footprint in the classification of Whole-Slide histopathology images since their training time is significantly reduced compared to the traditional CNN on the SQUH and BreaKHis datasets.

Subspace Model-Assisted Deep Learning for Improved Image Reconstruction.

Image reconstruction from limited and/or sparse data is known to be an ill-posed problem and a priori information/constraints have played an important role in solving the problem. Early constrained image reconstruction methods utilize image priors based on general image properties such as sparsity, low-rank structures, spatial support bound, etc. Recent deep learning-based reconstruction methods promise to produce even higher quality reconstructions by utilizing more specific image priors learned from training data. However, learning high-dimensional image priors requires huge amounts of training data that are currently not available in medical imaging applications. As a result, deep learning-based reconstructions often suffer from two known practical issues: a) sensitivity to data perturbations (e.g., changes in data sampling scheme), and b) limited generalization capability (e.g., biased reconstruction of lesions). This paper proposes a new method to address these issues. The proposed method synergistically integrates model-based and data-driven learning in three key components. The first component uses the linear vector space framework to capture global dependence of image features; the second exploits a deep network to learn the mapping from a linear vector space to a nonlinear manifold; the third is an unrolling-based deep network that captures local residual features with the aid of a sparsity model. The proposed method has been evaluated with magnetic resonance imaging data, demonstrating improved reconstruction in the presence of data perturbation and/or novel image features. The method may enhance the practical utility of deep learning-based image reconstruction.

Training Sample Pattern Optimization Based on a Swarm Intelligence Algorithm for Tiltrotor Flight Dynamics Model Approximation

Neural networks have been widely used as compensational models for aircraft control designs and as surrogate models for other optimizations. In the case of tiltrotor aircraft, the total number of aircraft states and controls is much greater than that of both traditional fixed-wings and helicopters. This requires, in general, a huge amount of training data for the network to reach a satisfactory approximation precision and makes the network size rise considerably. To solve the practical problem of reducing the size of the approximating network, efforts have to be made in the efficient utilization of the limited amount of training data. This work presents the methodology of optimizing the sample pattern of the training data set by adopting the metaheuristic algorithm of the particle swarm optimizer improved by the fourth-order Runge–Kutta algorithm. A 6-degree-of-freedom nonlinear flight dynamics model of the tiltrotor aircraft is derived, along with its approximation radial basis function neural network. An example case of approximating a highly nonlinear function is studied to illustrate the principle and main parameters of the optimizer, and the approximation performance of the time-domain response of the unstable nonlinear system is revealed by the study of a Van der Pol oscillator. Then, the presented method is applied to the modeled tiltrotor aircraft for its early and late conversion modes. The optimization scheme shows great improvement in both cases, as the function approximation error is reduced significantly.

Deep Random Forest and AraBert for Hate Speech Detection from Arabic Tweets

Nowadays, hate speech detection from Arabic tweets attracts the attention of many researchers. Numerous systems and techniques have been proposed to address this classification challenge. Nonetheless, three major limits persist: the use of deep learning models with an excess of hyperparameters, the reliance on hand-crafted features, and the requirement for a huge amount of training data to achieve satisfactory performance. In this study, we propose Contextual Deep Random Forest (CDRF), a hate speech detection approach that combines contextual embedding and Deep Random Forest. From the experimental findings, the Arabic contextual embedding model proves to be highly effective in hate speech detection, outperforming the static embedding models. Additionally, we prove that the proposed CDRF significantly enhances the performance of Arabic hate speech classification.

A benchmark dataset for deep learning-based airplane detection: HRPlanes

Airplane detection from satellite imagery is a challenging task due to the complex backgrounds in the images and differences in data acquisition conditions caused by the sensor geometry and atmospheric effects. Deep learning methods provide reliable and accurate solutions for automatic detection of airplanes; however, huge amount of training data is required to obtain promising results. In this study, we create a novel airplane detection dataset called High Resolution Planes (HRPlanes) by using images from Google Earth (GE) and labeling the bounding box of each plane on the images. HRPlanes include GE images of several different airports across the world to represent a variety of landscape, seasonal and satellite geometry conditions obtained from different satellites. We evaluated our dataset with two widely used object detection methods namely YOLOv4 and Faster R-CNN. Our preliminary results show that the proposed dataset can be a valuable data source and benchmark data set for future applications. Moreover, proposed architectures and results of this study could be used for transfer learning of different datasets and models for airplane detection.

Optimized Binaural Enhancement via attention masking network-based speech separation framework in digital hearing aids

Earlier studies have utilized microphone signal processing for performing the target speech evaluation and separation-like feature recognition by involving a huge amount of training data along with supervised machine learning. These approaches are most appropriate for stationary noise suppression; however, it is challenging for non-stationary noise, and it does not satisfy the practical processing requirement. To overcome the former issue, a system is presented for “joint speaker separation, and noise suppression”, referred to as the Optimized Binaural Enhancement via Attention Masking Network (OBEAMNET). In this paper, a speech separation model for a hearing aid is suggested. Here, the noise speech is separated into two different speeches through the proposed OBEAMNET approach. Initially, the standard input audio signals are gathered from standard benchmark datasets. Further, the cepstral features are gathered from the input data for getting the essential information. A novel Fitness Ordered Black Widow Optimization (FO-BWO) algorithm is developed for optimizing the features to increase the PSNR as well as decrease the RMSE. The feature selection is carried out by FO-BWO to get the most noteworthy information from audio signals. The suggested FO-BWO, along with the OBEAMNET technique, is used for efficiently separating the signals. Finally, the optimal features are fed to the speech separation using the OBEAMNET approach for improving the effectiveness of the proposed model that is observed in decoding auditory attention in noisy practical environments.

LORE: Logical Location Regression Network for Table Structure Recognition

Table structure recognition (TSR) aims at extracting tables in images into machine-understandable formats. Recent methods solve this problem by predicting the adjacency relations of detected cell boxes, or learning to generate the corresponding markup sequences from the table images. However, they either count on additional heuristic rules to recover the table structures, or require a huge amount of training data and time-consuming sequential decoders. In this paper, we propose an alternative paradigm. We model TSR as a logical location regression problem and propose a new TSR framework called LORE, standing for LOgical location REgression network, which for the first time combines logical location regression together with spatial location regression of table cells. Our proposed LORE is conceptually simpler, easier to train and more accurate than previous TSR models of other paradigms. Experiments on standard benchmarks demonstrate that LORE consistently outperforms prior arts. Code is available at https:// github.com/AlibabaResearch/AdvancedLiterateMachinery/tree/main/DocumentUnderstanding/LORE-TSR.

A Regularized Regression Thermal Error Modeling Method for CNC Machine Tools under Different Ambient Temperatures and Spindle Speeds

Establishing a mathematical model to predict and compensate for the thermal error of CNC machine tools is a commonly used approach. Most existing methods, especially those based on deep learning algorithms, have complicated models that need huge amounts of training data and lack interpretability. Therefore, this paper proposes a regularized regression algorithm for thermal error modeling, which has a simple structure that can be easily implemented in practice and has good interpretability. In addition, automatic temperature-sensitive variable selection is realized. Specifically, the least absolute regression method combined with two regularization techniques is used to establish the thermal error prediction model. The prediction effects are compared with state-of-the-art algorithms, including deep-learning-based algorithms. Comparison of the results shows that the proposed method has the best prediction accuracy and robustness. Finally, compensation experiments with the established model are conducted and prove the effectiveness of the proposed modeling method.

A new model based on multi-aspect images and complex-valued neural network for synthetic aperture radar automatic target recognition

ABSTRACT During the last decades, Synthetic Aperture Radar Automatic Target Recognition (SAR ATR) has been vastly utilized in the military services and civil studies. Because of the sensibility of the SAR images of the imaging in the azimuth dimension, using the multi-aspect SAR image sequence is more practical than the single-aspect SAR image to achieve the superior classification accuracy in an actual SAR ATR task. At present time, multi-aspect SAR ATR models mostly utilize Recurrent Neural Networks (RNN) that depend on the sequence between samples and so suffer from the lack of information. In a practical work, a huge amount of training set is needed to train a deep learning network accurately, but it is so costly work to extract multi-aspect SAR images. So, in this article, a new model of multi-aspect SAR ATR is proposed by using a self-attention model. The proposed self-attention model is utilized to calculate the internal correlation between the original SAR images. At the same time, to develop the anti-noise capability of the proposed model and decrease the dependency on a huge amount of training data, a Convolutional Auto Encoder (CAE) is designed and utilized in the feature extraction section of the proposed model. On the other hand, unlike existing traditional methods, in this work we use both amplitude and phase information of SAR images to devolve the training process of the proposed model. It should be noted that all of the parameters of the proposed network are developed to a complex domain. In addition a complex backpropagation algorithm by using the gradient based model is used for training the network. In the end, experiments is obtained by using two MSTAR data set (the MSTAR-SOC and EOC). The experimental results prove that the proposed model not only can achieve a high recognitions rate on the case of sufficient training sample but also can obtain an acceptable rate in the case of small training samples in different complex situations. In addition, the simulation results demonstrate that by using the encoder of CAE in the proposed model, the whole configuration of the proposed model achieves the anti-noise capability that is a valuable benefit of any practical SAR ATR task.

Classification of SARS-CoV-2 viral genome sequences using Neurochaos Learning.

The high spread rate of SARS-CoV-2 virus has put the researchers all over the world in a demanding situation. The need of the hour is to develop novel learning algorithms that can effectively learn a general pattern by training with fewer genome sequences of coronavirus. Learning from very few training samples is necessary and important during the beginning of a disease outbreak when sequencing data is limited. This is because a successful detection and isolation of patients can curb the spread of the virus. However, this poses a huge challenge for machine learning and deep learning algorithms as they require huge amounts of training data to learn the pattern and distinguish from other closely related viruses. In this paper, we propose a new paradigm – Neurochaos Learning (NL) for classification of coronavirus genome sequence that addresses this specific problem. NL is inspired from the empirical evidence of chaos and non-linearity at the level of neurons in biological neural networks. The average sensitivity, specificity and accuracy for NL are 0.998, 0.999 and 0.998 respectively for the multiclass classification problem (SARS-CoV-2, Coronaviridae, Metapneumovirus, Rhinovirus and Influenza) using leave one out crossvalidation. With just one training sample per class for 1000 independent random trials of training, we report an average macro F1-score > 0.99 for the classification of SARS-CoV-2 from SARS-CoV-1 genome sequences. We compare the performance of NL with K-nearest neighbours (KNN), logistic regression, random forest, SVM, and naïve Bayes classifiers. We foresee promising future applications in genome classification using NL with novel combinations of chaotic feature engineering and other machine learning algorithms.Graphical abstract Supplementary InformationThe online version contains supplementary material available at 10.1007/s11517-022-02591-3.

CrossViT Wide Residual Squeeze-and-Excitation Network for Alzheimer’s disease classification with self attention ProGAN data augmentation

Efficient and accurate early prediction of Alzheimer’s disease (AD) based on the neuroimaging data has attracted interest from many researchers to prevent its progression. Deep learning networks have demonstrated an optimal ability to analyse large-scale multimodal neuroimaging for AD classification. The most widely used architecture of deep learning is the Convolution neural networks (CNN) that have shown great potential in AD detection. However CNN does not capture long range dependencies within the input image and does not ensure a good global feature extraction. Furthermore, increasing the receptive field of CNN by increasing the kernels sizes can cause a feature granularity loss. Another limitation is that CNN lacks a weighing mechanism of image features; the network doesn’t focus on the relevant features within the image. Recently,vision transformer have shown an outstanding performance over the CNN and overcomes its main limitations. The vision transformer relies on the self-attention layers. The main drawbacks of this new technique is that it requires a huge amount of training data. In this paper, we combined the main strengths of these two architectures for AD classification. We proposed a new method based on the combination of the Cross ViT and Wide Residual Squeeze-and-Excitation Network. We acquired MRI data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Open Access Series of Imaging Studies (OASIS). We also proposed a new data augmentation based on the self attention progressive generative adversarial neural network to overcome the limitation of the data. Our proposed method achieved 99% classification accuracy and outperforms CNN models.

Detection of bladder cancer with feature fusion, transfer learning and CapsNets

This paper confronts two approaches to classify bladder lesions shown in white light cystoscopy images when using small datasets: the classical one, where handcrafted-based features feed pattern recognition systems and the modern deep learning-based (DL) approach. In between, there are alternative DL models that had not received wide attention from the scientific community, even though they can be more appropriate for small datasets such as the human brain motivated capsule neural networks (CapsNets). However, CapsNets have not yet matured hence presenting lower performances than the most classic DL models. These models require higher computational resources, more computational skills from the physician and are more prone to overfitting, making them sometimes prohibitive in the routine of clinical practice. This paper shows that carefully handcrafted features used with more robust models can reach similar performances to the conventional DL-based models and deep CapsNets, making them more useful for clinical applications. Concerning feature extraction, it is proposed a new feature fusion approach for Ta and T1 bladder tumor detection by using decision fusion from multiple classifiers in a scheme known as stacking of classifiers. Three Neural Networks perform classification on three different feature sets, namely: Covariance of Color Histogram of Oriented Gradients, proposed in the ambit of this paper; Local Binary Patterns and Wavelet Coefficients taken from lower scales. Data diversity is ensured by a fourth Neural Network, which is used for decision fusion by combining the outputs of the ensemble elements to produce the classifier output. Both Feed Forward Neural Networks and Radial Basis Functions are used in the experiments. Contrarily, DL-based models extract automatically the best features at the cost of requiring huge amounts of training data, which in turn can be alleviated by using the Transfer Learning (TL) strategy. In this paper VGG16 and ResNet-34 pretrained in ImageNet were used for TL, slightly outperforming the proposed ensemble. CapsNets may overcome CNNs given their ability to deal with objects rotational invariance and spatial relationships. Therefore, they can be trained from scratch in applications using small amounts of data, which was beneficial for the current case, improving accuracy from 94.6% to 96.9%.

An Ancient Number Recognition using Freeman Chain Code with Deep Learning Approach

Sanskrit character and number documents have a lot of errors. Correcting those errors using conventional spell-checking approaches breaks down due to the limited vocabulary. This is because of high inflexions of Sanskrit, where words are dynamically formed by Sandhi rules, Samasa rules, Taddhita affixes, etc. Therefore, correcting OCR documents require huge efforts. Here, we can present different machine learning approaches and various ways to improve features for ameliorating the error corrections in Sanskrit documents. Simulation of Sanskrit dictionary for synthesizing off-the-shelf dictionary can be done. Most of the proposed methods can also work for general Sanskrit word corrections and Hindi word corrections. Handwriting recognition in Indic scripts, like Devanagari, is very challenging due to the subtitles in the scripts, variations in rendering and the cursive nature of the handwriting. Lack of public handwriting datasets in Indic scripts has long stymied the development of offline handwritten word recognizers and made comparison across different methods a tedious task in the field. In this paper, a new handwritten word dataset will be released for Devanagari, IIIT-HW-Dev to alleviate some of these issues. This process is required for successful training of deep learning architecture, availability of huge amounts of training data is crucial, as any typical architecture contains millions of parameters. A new method for the classification of freeman chain code using four-connectivity and eight-connectivity events with deep learning approach is presented. Application of CNN LeNet-5 is found to be suitable to get results in this cases as the numbers are formed with curved lines In contrast with the existing FCC event data analysis techniques, sampled grey images of the existing events are not used, but image files of the three-phase PQ event data are analysed by taking the advantage of the success of the deep learning approach on imagefile-classification. Therefore, the novelty of the proposed approach is that image files of the voltage waveforms of the three phases of the power grid are classified. It is shown that the test data can be classified with 100% accuracy. The proposed work is believed to serve the needs of the future smart grid applications, which are fast and taking automatic countermeasures against potential PQ events.

Accuracy Assessment of Joint Angles Estimated from 2D and 3D Camera Measurements.

To automatically evaluate the ergonomics of workers, 3D skeletons are needed. Most ergonomic assessment methods, like REBA, are based on the different 3D joint angles. Thanks to the huge amount of training data, 2D skeleton detectors have become very accurate. In this work, we test three methods to calculate 3D skeletons from 2D detections: using the depth from a single RealSense range camera, triangulating the joints using multiple cameras, and combining the triangulation of multiple camera pairs. We tested the methods using recordings of a person doing different assembly tasks. We compared the resulting joint angles to the ground truth of a VICON marker-based tracking system. The resulting RMS angle error for the triangulation methods is between 12° and 16°, showing that they are accurate enough to calculate a useful ergonomic score from.

Rapid Quantitative Analysis of IR Absorption Spectra for Trace Gas Detection by Artificial Neural Networks Trained with Synthetic Data.

Infrared absorption spectroscopy is a widely used tool to quantify and monitor compositions of gases. The concentration information is often retrieved by fitting absorption profiles to the acquired spectra, utilizing spectroscopic databases. In complex gas matrices an expanded parameter space leads to long computation times of the fitting routines due to the increased number of spectral features that need to be computed for each iteration during the fit. This hinders the capability of real-time analysis of the gas matrix. Here, an artificial neural network (ANN) is employed for rapid prediction of gas concentrations in complex infrared absorption spectra composed of mixtures of CO and N2O. Experimental data is acquired with a mid-infrared dual frequency comb spectrometer. To circumvent the experimental collection of huge amounts of training data, the network is trained on synthetically generated spectra. The spectra are based on simulated absorption profiles making use of the HITRAN database. In addition, the spectrometer’s influence on the measured spectra is characterized and included in the synthetic training data generation. The ANN was tested on measured spectra and compared to a non-linear least squares fitting algorithm. An average evaluation time of 303 µs for a single measured spectrum was achieved. Coefficients of determination were 0.99997 for the predictions of N2O concentrations and 0.99987 for the predictions of CO concentrations, with uncertainties on the predicted concentrations between 0.04 and 0.18 ppm for 0 to 100 ppm N2O and between 0.05 and 0.18 ppm for 0 to 60 ppm CO.

Power System Coherency Detection From Wide-Area Measurements by Typicality-Based Data Analysis

This paper presents a new data-driven methodology for power system coherency identification of generator and non-generator buses. This methodology is exclusively based on intrinsic statistical properties extracted directly from observations, without any prior assumption of the probability distribution function (PDF) for the data. The main advances of this proposal are: ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i</i> ) gathering of statistical information from the data itself despite scenarios where the PDF may change (different inverter-based load and generation scenarios, load levels of the system, and changes in topology); and ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ii</i> ) assignment of buses into coherent areas without any tuning of parameters, nor manually labeling of huge amounts of training data. This new method, called typicality-based data analysis (TDA), is applied to the correlation metric of the distance between dynamic responses of buses, either voltage angles or frequencies. Simulated signals from a benchmark power system with cases considering the presence of non-synchronous generation and islanding conditions, and real data associated with generation trips in the U.S. Eastern Interconnection are used to corroborate the methodology effectiveness.

Real-time object detector for the visually impaired with voice feedback using openCV

The goal of this paper is to create an object detector model that can detect objects for visually impaired people and other commercial users by detecting it at a certain distance. Existing object detection algorithms required a huge amount of training data, which took longer and was extremely complex. This is also a difficult task. As a result, it presents a computer vision paradigm for converting an object to text by importing a pre-trained CAFFEMODEL (a machine learning model created by Caffe) framework dataset model, and the texts are further converted to speech. This method allows the detection of multiple objects on the same screen. It helps in real-time object detection. This paper discusses the concept, methodology, and system architecture for the implementation of the system in combination with the obtained intermediate results and analyzes the tools used in the proposed system. This system can then be implemented in any other system. Portable gadgets that detect objects at a certain distance from visually impaired people and transmit a voice signal.

An Accurate Real-Time License Plate Detection Method Based On Deep Learning Approaches

In vision-driven Intelligent Transportation Systems (ITS) where cameras play a vital role, accurate detection and re-identification of vehicles are fundamental demands. Hence, recent approaches have employed a wide range of algorithms to provide the best possible accuracy. These methods commonly generate a vehicle detection model based on its visual appearance features such as license plate, headlights, or some other distinguishable specifications. Among different object detection approaches, Deep Neural Networks (DNNs) have the advantage of magnificent detection accuracy in case a huge amount of training data is provided. In this paper, a robust approach for license plate detection (LPD) based on YOLO v.3 is proposed which takes advantage of high detection accuracy and real-time performance. The mentioned approach can detect the license plate location of vehicles as a general representation of vehicle presence in images. To train the model, a dataset of vehicle images with Iranian license plates has been generated by the authors and augmented to provide a wider range of data for test and train purposes. It should be mentioned that the proposed method can detect the license plate area as an indicator of vehicle presence with no Optical Character Recognition (OCR) algorithm to distinguish characters inside the license plate. Experimental results have shown the high performance of the system with a precision 0.979 and recall 0.972.