Histogram of oriented gradients

Homicide crime prediction is a critical area of focus for law enforcement agencies aiming to enhance public safety and efficiently allocate resources. This study develops and evaluates several machine learning models to predict homicide crimes using historical data from 2007 to 2022. The models assessed include Random Forest, Decision Tree, Logistic Regression, and Linear Regression. Performance evaluation reveals that the Decision Tree model achieved the highest accuracy at 99%, with a precision of 0.99, recall of 0.98, and F1 score of 0.99. Logistic Regression also performed well, with an accuracy of 95%, precision of 0.94, recall of 0.96, and F1 score of 0.95. These results demonstrate the potential of machine learning models in predicting homicide crimes. A web application was developed to make these predictions accessible and actionable for law enforcement agencies, aiding in resource allocation and crime prevention strategies.

The iris is considered as the biometric trait with the highest unique probability. The iris location is an important task for biometrics systems, affecting directly the results obtained in specific applications such as iris recognition, spoofing and contact lenses detection, among others. This work defines the iris location problem as the delimitation of the smallest squared window that encompasses the iris region. In order to build a benchmark for iris location we annotate (iris squared bounding boxes) four databases from different biometric applications and make them publicly available to the community. Besides these 4 annotated databases, we include 2 others from the literature. We perform experiments on these six databases, five obtained with near infra-red sensors and one with visible light sensor. We compare the classical and outstanding Daugman iris location approach with two window based detectors: 1) a sliding window detector based on features from Histogram of Oriented Gradients (HOG) and a linear Support Vector Machines (SVM) classifier; 2) a deep learning based detector fine-tuned from YOLO object detector. Experimental results showed that the deep learning based detector outperforms the other ones in terms of accuracy and runtime (GPUs version) and should be chosen whenever possible.

As known from the literature, machine learning (ML) is one of the popular researches have been used variable areas. In this work, a novel exemplar pyramid method is presented to accurately classify Astragalus L. taxa by using their chromosome images. To implement ML to biological images, the proposed exemplar pyramid method is used. Histogram of Oriented Gradients (HOG) is utilized as feature generator. The proposed exemplar pyramid method consists of preprocessing, feature generation and concatenation, feature selection and classification phase. 10 classifiers are chosen to train and test the extracted features. According to results, the proposed exemplar pyramid generates discriminative features. because five of the used 10 classifiers achieved 100.0% classification rate.

We propose a method capable to predict vehicle trajectories in a real scenario based on an unsupervised approach using Histogram of Oriented Gradients (HOG) features to construct an uniform path. The proposed algorithm extracts a sub-region of the input image defined as Field of View of the target vehicle, to output a possible trajectory that the given vehicle will follow through. We perform many experiments using the proposed technique, and based on qualitative/quantitative analyses, we conclude it is successfully able to predict reasonable trajectories.

Search and rescue is often time and labour intensive. We present a system to be used in drones to make search and rescue operations more effective. The system uses a drone downward facing camera to detect people and to evaluate potential sites as being safe or not for the drone to land. Histogram of Oriented Gradients (HOG) features are extracted and a Support Vector Machine (SVM) is used as classifier. Our results show good performance on classifying frames as containing people (Sensitivity > 78%, Specificity > 83%), and distinguishing between safe and dangerous landing sites (Sensitivity > 87%, Specificity > 98%).

With the rapidly increasing crime rate in recent years, community safety issues aroused a wide concern among public community. Various security technologies had been invented and carried out, for example password door lock, alarm system, and closed-circuit televisions (CCTVs). Although the installation of CCTVs is common in most premises, they require extensive man power to manually monitor the videos. Moreover, the reliability of human operator greatly deteriorates when they are in fatigue condition. In view of this, our project aims to develop an automated computer vision based surveillance system. Unlike ordinary CCTV system that requires human operator to manually observe and detect intruder, a computer vision based surveillance system automatically monitor the security of premises and trigger actions once an intrusion is detected. Basically, it is a simple surveillance camera system that will be setup at the entrance of the house. The reliability is being enhanced by applying the motion detection and face recognition algorithm, using histogram of oriented gradients that could detect the existence of people at the main entrance and try to validate the user. Apart from recognizing the user, the propose system also support mobile interaction whereby user can monitor the camera, activate alarm, and even received notification when a stranger was being detected at the entrance of the house. By including such functionalities, proposed system had highly surpassed the existing surveillance system by not only support monitoring, but also try to recognize the people and inform the user at the exact moment when stranger detected, so that user could take immediate action about it, for example activating the alarm or report to police. The project was executed with expected outcome and objectives had been accomplished.

Identifying a person primarily relies on their facial features, which even distinguish identical twins. As a result, facial recognition and identification become crucial for distinguishing individuals. Biometric authentication technology, specifically facial recognition systems, are utilized to verify one’s identity. This technology has gained popularity in modern applications, such as phone unlock systems, criminal identification systems, and home security systems. Due to its reliance on a facial image rather than external factors like a card or key, this method is considered more secure. The process of recognizing a person involves two primary steps: face detection and face identification. This article delves into the concept of developing a face recognition system utilizing Python’s OpenCV library through deep learning. Due to its exceptional accuracy, deep learning is an ideal method for facial recognition. The proposed approach involves utilizing the Haar cascade techniques for...

Verifying the authenticity of handwritten signatures is required in various current life domains, notably with official contracts, banking or financial transactions. Therefore, in this paper a novel histogrambased descriptor and an improved classification of the bio-inspired Artificial Immune Recognition System (AIRS) are proposed for handwritten signature verification. Precisely, the Histogram Of Templates (HOT) is introduced to characterize the most widespread orientations of local strokes in handwritten signatures, while the combination of AIRS and SVM is proposed to achieve the verification task. Usually, using the k Nearest Neighbor rule, a questioned signature is classified by computing dissimilarities with respect to all AIRS outputs. In this work, using these dissimilarities, a second round of training is achieved by the SVM classifier to further improve the discrimination power. In comparison with existing methods, the experiments on two widely-used datasets show the potential and the effectiveness of the proposed system.

This study suggests a Convolutional Neural Network (CNN) and Histogram Oriented Gradients (HOG)-based automatic detection and classification system for mango disease. Early detection is essential for efficient disease management since mango disease can have a major influence on fruit quality and yield. The suggested system makes use of the CNN algorithm for extracting features and the HOG technique for capturing shape and texture data. The extracted features are subsequently used to feed a disease classification model for disease detection. The efficiency of the proposed model is demonstrated by experimental findings, which achieve excellent accuracy in both disease detection and classification tasks. The CNN-HOG hybrid model outperforms CNN or HOG alone in terms of performance, demonstrating the complementary nature of these two methods for the detection and classification of mango disease. The system's performance is evaluated using measures for accuracy, precision, and recall and the accuracy of the proposed model is 98.80%. This research helps establish effective and trustworthy tools for managing mango disease by automating the detection and classification process. This enables prompt intervention and reduces crop losses.

In this work, we present an ensemble of descriptors for the classification of virus images acquired using transmission electron microscopy. We trained multiple support vector machines on different sets of features extracted from the data. We used both handcrafted algorithms and a pretrained deep neural network as feature extractors. The proposed fusion strongly boosts the performance obtained by each stand-alone approach, obtaining state of the art performance.

In recent years, active learning has emerged as a powerful tool in building robust systems for object detection using computer vision. Indeed, active learning approaches to on-road vehicle detection have achieved impressive results. While active learning approaches for object detection have been explored and presented in the literature, few studies have been performed to comparatively assess costs and merits. In this study, we provide a cost-sensitive analysis of three popular active learning methods for on-road vehicle detection. The generality of active learning findings is demonstrated via learning experiments performed with detectors based on histogram of oriented gradient features and SVM classification (HOG-SVM), and Haar-like features and Adaboost classification (Haar-Adaboost). Experimental evaluation has been performed on static images and real-world on-road vehicle datasets. Learning approaches are assessed in terms of the time spent annotating, data required, recall, and precision.

Handwritten signatures remain a widely accepted form of identity verification in financial, legal, governmental, and institutional settings. However, they are highly vulnerable to forgery, particularly in offline scenarios where dynamic behavioral traits like pen pressure, speed, and stroke order are unavailable. Traditional methods, including manual verification by forensic experts and rule-based automated systems, suffer from subjectivity, limited scalability, and poor performance against skilled forgeries. This research presents a Convolutional Neural Network (CNN)-based framework for offline signature forgery detection that leverages deep learning to extract discriminative features directly from static signature images. The proposed system incorporates robust preprocessing techniques, advanced data augmentation to address class imbalance, and transfer learning using state-of-the-art architectures like ResNet, VGG, EfficientNet, and MobileNetV2. To improve transparency and legal applicability, the system integrates visual interpretability tools such as Grad-CAM and saliency maps to highlight regions influencing model decisions. Experiments conducted on five benchmark datasets-GPDS, CEDAR, BHSig260, UTSig, and MCYTdemonstrate the superior performance of the proposed model over traditional machine learning approaches. The CNN-based system achieved accuracy above 96% on all datasets, with Equal Error Rates (EER) as low as 1.7%. It also supports mobile and real-time deployment, making it viable for use in banking kiosks, forensic tools, and digital identity platforms.

Object detection is a big challenge for researchers to address the issues that affect accurate detections. The Histogram of oriented gradients (HOG) descriptors have been used extensively for object detection on challenging conditions with good results. However, occlusion remains a well-known issue where some parts of an object are only visible. Conventional frameworks using HOG descriptors for detecting whole human body are prone to occluded people and thus causing errors on implementations running these frameworks. This issue was addressed in different approaches like hybrid feature extraction algorithms. Rather than following such approaches, the problem is addressed in this research by redefining HOG descriptors to work in a part object (body) detection framework that is powered by a finite-state machine. Experiments were conducted to test a new method of extracting HOG descriptors for parts of human body via slicing a window’s HOG descriptor into four. The part body detection f...

Implementing accurate and reliable passenger detection and counting system is an important task for the correct distribution of available transport system. The aim of this paper is to develop an accurate computer vision-based system to track and count passengers. The proposed passenger detection system incorporates the ideas of well-established detection techniques and is optimally customised for both indoor and outdoor scenarios. The candidate foreground regions (inside an image) are extracted in the proposed method and are described using the histograms of oriented gradient descriptor. These features are trained and tested using support vector machine classifier and the detected passengers are tracked using a filter. The proposed counting system is used to count passengers automatically when they pass through a virtual line of interest. Accuracies ranging 91.2 percent to 86.24 percent were found for passenger detection using the proposed passenger detection and counting system whe...

Biological vision incorporates intelligent cooperation between the sensory and the motor systems, which is facilitated by the development of motor skills that help to shape visual information that is relevant to a specific vision task. In this article, we seek to explore an approach to active vision inspired by biological systems, which uses limited constraints for motor strategies through progressive adaptation via an evolutionary method. This type of approach gives freedom to artificial systems in the discovery of eye-movement strategies that may be useful to solve a given vision task but are not known to us. In the experiment sections of this article, we use this type of evolutionary active vision system for more complex natural images in both two-dimensional (2D) and three-dimensional (3D) environments. To further improve the results, we experiment with the use of pre-processing the visual input with both the uniform local binary patterns (ULBP) and the histogram of oriented gra...

The main component for head recognition is a feature extraction. One of them as our novel method is histogram of transition. In this paper we evaluate multi orientation performance of this feature for human head detection. The input images are head and shoulder image with angle of 315 o , 330 o , 345 o , 15 o , 30 o and 45 o . We use SVM classifier to recognize the input image as a head or non head, which is trained by using normal orientation (0 o ) images. For comparison, we compare the recognition rate with the existing method of feature extraction, i.e. Histogram of Oriented Gradient (HOG) and Linear Binary Pattern (LBP). The experimental results show our feature more robust than the existing feature.

This paper describes the detection of coconut trees using very-high-resolution optical satellite imagery. The satellite imagery used in this study was a panchromatic band of Pleiades imagery with a spatial resolution of 0.5 metres. The authors proposed the use of a histogram of oriented gradients (HOG) algorithm as the feature extractor and a support vector machine (SVM) as the classifier for this detection. The main objective of this study is to find out the parameter combination for the HOG algorithm that could provide the best performance for coconut-tree detection. The study shows that the best parameter combination for the HOG algorithm is a configuration of 3 x 3 blocks, 9 orientation bins, and L2-norm block normalization. These parameters provide overall accuracy, precision and recall of approximately 80%, 73% and 87%, respectively.

In this proposed work, the moving object is localized using curvelet transform, soft thresholding and frame differencing. The feature extraction techniques are applied on to the localized object and the texture, color and shape information of objects are considered. To extract the shape information, Speeded Up Robust Features (SURF) is used. To extract the texture features, the Enhanced Local Vector Pattern (ELVP) and to extract color features, Histogram of Gradient (HOG) are used and then reduced feature set obtained using genetic algorithm are fused to form a single feature vector and given into the Extreme Learning Machine (ELM) to classify the objects. The performance of the proposed work is compared with Naive Bayes, Support Vector Machine, Feed Forward Neural Network and Probabilistic Neural Network and inferred that the proposed method performs better.

Human face is considered as one of the most useful traits in biometrics, and it has been widely used in education, security, military and many other applications. However, in most of currently deployed face recognition systems ideal imaging conditions are assumed; to capture a fully featured images with enough quality to perform the recognition process. As the unmasked face will have a considerable impact on the numbers of new infections in the era of COVID-19 pandemic, a new unconstrained partial facial recognition method must be developed. In this research we proposed a mask detection method based on HOG (Histogram of Gradient) features descriptor and SVM (Support Vector Machine) to determine whether the face is masked or not, the proposed method was tested over 10000 randomly selected images from Masked Face-Net database and was able to correctly classify 98.73% of the tested images. Moreover, and to extract enough features from partially occluded face images, a new geometrical f...

We propose using Immersive Virtual Reality activities to improve the spatial ability of engineering students based on the study of solid geometry. The work group is selected randomly from among all the students registered for the 1st term course Graphic Expression and Computer-Aided Design (GECAD) at the Barcelona College of Industrial Engineering (EUETIB). A total of 60 participants completed three activities (6 h) in VR, using head-mounted display (HMD) glasses. Another group of students (30) made up the control group, which carried out only the learning activities that were common to all students, in a SolidWorks 3D non-immersive solid modeling software environment. Spatial abilities are assessed using the Differential Aptitude Test: Spatial Relations Subset (DAT:SR) and the Purdue Spatial Visualization Test: Rotations (PSVT:R). Previous studies have demonstrated a close correlation between successful comprehension of the Graphic Engineering course contents and high scores on the...

The authors propose a novel-makeup detection approach that assists face recognition systems to achieve a higheraccuracy rate while dealing with makeup images. Makeup features are defined in this work using biologically inspired features (BIFs). To establish makeup depictive features more specifically, colour and texture features are essential to be extracted from images. Hence, they create makeup depictive features in the last complex layer of BIFs (C2) as average skin tone (AST) and a histogram of oriented gradient (HOG), where AST is representative of colour and HOG exhibits texture. The proposed makeup BIFs are extracted from grayscale images and instead of breaking the face image into several patches, the whole face image is employed. This resulted in the performance acceleration as well as higher accuracy rate compared with state-of-the-art makeup detection schemes. Subsequently, they employ a machine learning scheme to train the makeup detection system by feeding it makeup and non-makeup labelled images. They exploit the correlation-based method for the face recognition system and compare the results with the direct two-dimensional principal component analysis face recognition scheme for makeup datasets. Experimental results show the highest accuracy rate of 97.07% was achieved by the proposed algorithm for face recognition system considering makeup.

Computer vision enables a wide range of applications in robotics/drones, self-driving cars, smart Internet of Things, and portable/wearable electronics. For many of these applications, local embedded processing is preferred due to privacy and/or latency concerns. Accordingly, energy-efficient embedded vision hardware delivering real-time and robust performance is crucial. While deep learning is gaining popularity in several computer vision algorithms, a significant energy consumption difference exists compared to traditional hand-crafted approaches. In this paper, we provide an in-depth analysis of the computation, energy and accuracy trade-offs between learned features such as deep Convolutional Neural Networks (CNN) and hand-crafted features such as Histogram of Oriented Gradients (HOG). This analysis is supported by measurements from two chips that implement these algorithms. Our goal is to understand the source of the energy discrepancy between the two approaches and to provide insight about the potential areas where CNNs can be improved and eventually approach the energy-efficiency of HOG while maintaining its outstanding performance accuracy.

Due to the advancements in digital technologies and social networking, image collections are growing exponentially. The important aim in content-based image retrieval (CBIR) is to reduce the semantic gap issue that improves the performance of image retrieval. In this paper, the objective is achieved by introducing effective visual words fusion technique based on speeded-up robust features (SURF) and histograms of oriented gradients (HOG) feature descriptors. HOG is used to extract global features, whereas SURF is used for the extraction of local features. Global features are preferred for large-scale image retrieval, whereas local features perform better on those systems that support semantic queries with close visual appearance. Moreover, SURF is scale and rotation-invariant as compared to HOG descriptor and it works better for low illumination. On the contrary, HOG performs better for scene-recognition-or activity-recognition-based applications. In the proposed technique, visual words fusion of SURF and HOG feature descriptors is carried which performed better than features fusion of SURF and HOG feature descriptors as well as state-of-the-art CBIR techniques. The proposed technique based on visual words fusion gives classification accuracy of 98.40% using support vector machine while image retrieval accuracy of 80.61%. Qualitative and quantitative analyses performed on four standard image collections namely, Corel-1000, Corel-1500, Corel-5000, and Caltech-256 demonstrate the effectiveness of the proposed technique based on visual words fusion of SURF and HOG feature descriptors. Content-based image retrieval • Classification • Local and global features • Semantic gap • Visual words fusion B Zahid Mehmood

Traffic accidents are caused by several factors, especially due to driver fatigue. To minimize accidents caused by human negligence, developing a prototype microsleep detection system to trigger an alarm is necessary. This research uses Histogram Oriented Gradient and Support Vector Machine methods to detect objects. The programming language used is Python, using the dataset from the Idlib face landmark as a marker in the face point area, which will then be calculated based on the eye aspect ratio. This system is implemented using video input captured using a webcam in real-time. The output of this system uses a buzzer to alert the driver. In this study, the test results were obtained well, carried out with 2 test scenarios with a distance of 40cm - 100cm and testing light levels of 33 lux to 226 lux. From these results, the accuracy results were obtained at 88% each.

Most reported works on fingertip detection focus on extended fingers where the hand is not occluded by another object. This paper proposes a machine-vision-based technique exploiting the contour of the hand and fingers for detecting the fingertips when the hand is manipulating a ball, which means that the fingers are closed and the hand is partially occluded. The preliminary result of our on-going research is promising where it can be used to generate a more objective performance indicator for monitoring the progress during hand therapy by using a digital webcam. Being markerless and contactless, the proposed technique will require minimal preparation prior to the therapy.

Human action recognition involves recognizing and classifying actions performed by humans. It has many applications, including sports, healthcare, and surveillance. Challenges such as a limited number of classes of activities and variations within inter and intra-class groups lead to high misclassification rates in some of the intelligent systems developed. Existing studies focused mainly on using public datasets with little focus on real-life action datasets, with limited research on HAR for military obstacle-crossing activities. This paper focuses on recognizing human actions in an obstaclecrossing competition video sequence where multiple participants are performing different obstaclecrossing activities. This study proposes a feature descriptor approach that combines a Histogram of Oriented Gradient and Region Descriptors (HOGReG) for human action recognition in a military obstacle crossing competition. The dataset was captured during military trainees' obstacle-crossing exercises at a military training institution to achieve this objective. Images were segmented into background and foreground using a Grabcut-based segmentation algorithm, and thereafter, features were extracted and used for classification. The features were extracted using a Histogram of Oriented Gradient (HOG) and region descriptors from segmented images. The extracted features are presented to a neural network classifier for classification and evaluation. The experimental results recorded 63.8%, 82.6%, and 86.4% recognition accuracies using the region descriptors HOG and HOGReG, respectively. The region descriptor gave a training time of 5.6048 seconds, while HOG and HOGReG reported 32.233 and 31.975 seconds, respectively. The outcome shows how effectively the suggested model performed.

Radon transform and its inverse operation are important techniques in medical imaging tasks. Recently, there has been renewed interest in Radon transform for applications such as content-based medical image retrieval. However, all studies so far have used Radon transform as a global or quasiglobal image descriptor by extracting projections of the whole image or large sub-images. This paper attempts to show that the dense sampling to generate the histogram of local Radon projections has a much higher discrimination capability than the global one. In this paper, we introduce Local Radon Descriptor (LRD) and apply it to the IRMA dataset, which contains 14,410 xray images as well as to the INRIA Holidays dataset with 1,990 images. Our results show significant improvement in retrieval performance by using LRD versus its global version. We also demonstrate that LRD can deliver results comparable to wellestablished descriptors like LBP and HOG.

Computer vision applications face various challenges while detection and classification of objects in real world like large variation in appearances, cluttered back ground, noise, occlusion, low illumination etc.. In this paper a Wavelet based Histogram of Oriented Gradients (WHOG) feature descriptors are proposed to represent shape information by storing local gradients in image. This results in enhanced representation of shape information. The performance of the feature descriptors are tested on multiclass image data set having partial occlusion, different scales and rotated object images. The performance of WHOG feature based object classification is compared with HOG feature based classification. The matching of test image with its learned class is performed using Back Propagation Neural Network (BPNN) algorithm. Proposed features not only performed superior than HOG but also beat wavelet, moment invariant and Curvelet.

This paper presents a technique to classify the circular traffic sign based-on HOG (histogram of oriented gradients) and a ring partitioned matching. The method divides an image into several ring areas, and calculates the HOG feature on each ring area. In the matching process, the weight is assigned to each ring for calculating the distance of HOG feature between tested image and reference image. The experimental results show that the proposed algorithm achieves a high classification rate of 97.8%, without the need of many prepared sample images. The results also show that the best values of the number of orientation bins and the cell size of the HOG parameters are 5 and 10 x 10 pixels respectively.

This paper presents a technique to classify the circular traffic sign based-on HOG (histogram of oriented gradients) and a ring partitioned matching. The method divides an image into several ring areas, and calculates the HOG feature on each ring area. In the matching process, the weight is assigned to each ring for calculating the distance of HOG feature between tested image and reference image. The experimental results show that the proposed algorithm achieves a high classification rate of 97.8%, without the need of many prepared sample images. The results also show that the best values of the number of orientation bins and the cell size of the HOG parameters are 5 and 10 x 10 pixels respectively.

The world we live in today has an utmost importance to have a video surveillance system for detecting any kind of violent behaviours, for example, airports, railway stations, etc. In the not so distant past, the rate of violence has increased drastically. However, the traditional violence detection system uses low-level feature extraction along with other Sapio-Temporal features models developed before in order to extract high-level features. The existing systems in place are able to detect the violent footage in videos using the traditional methods. These methods include motion regions segmentation as per the distribution of optical flow fields. This is done by using low-level features, extracted from RGB images, using the very well known "Local Histogram of Oriented Gradient", or LHOG and from optical flow images using the method of "Local Histogram of Optical Flow", or LHOF. Further in the process, the features that were extracted from the images are coded usi...

Histogram of Oriented Gradients (HOG) is an object detection algorithm used to detect people from an image. It involves features extraction called ‘HOG descriptor’ which are used to identify a person in the image. Several operations are involved in the feature extraction process. Hence performing numerous computations in order to obtain HOG descriptors takes some considerable amount of time. This slow computation speed limits HOG’s application in real-time systems. This paper investigates HOG with a view to improve its speed, modify the feature computation process to develop a faster version of HOG and finally evaluate against existing HOG. The technique of asymptotic notation in particular Big-O notation was applied to each stage of HOG and the complexity for the binning stage was modified. This results in a HOG version with a reduced complexity from n4 to n2 thereby having an improved speed as compared to the original HOG.

Image classification involves categorizing an image's pixels into specific classes based on their unique characteristics. It has diverse applications in everyday life. One such application is the classification of diseases on corn leaves. Corn is a widely consumed staple food in Indonesia, and healthy corn plants are crucial for meeting market demands. Currently, disease identification in corn plants relies on manual checks, which are time-consuming and less effective. This research aims to automate disease identification on corn leaves using the Support Vector Machine (SVM), K-Nearest Neighbor (K-NN) with K=2, and Stochastic Gradient Descent (SGD) algorithms. The classification process utilizes the Histogram of Oriented Gradients (HOG) feature extraction method with a dataset of corn leaf images. The classification results achieved an accuracy of 71.44%, AUC of 79.16%, precision of 70.08%, recall of 71.44%, and f1 score of 67.11%. The highest accuracy was obtained by combining ...

In the realm of digital marketing for the banking industry, the integration of deep learning methodologies, particularly
Convolutional Neural Networks (CNNs) such as VGG16, Resnet50, and InceptionV3, has revolutionized strategic decision-making
and customer satisfaction. This study explores how deep learning models leverage neural networks with multiple layers to
analyze vast and complex datasets, uncovering intricate patterns in customer behavior and preferences. By enhancing customer
segmentation, optimizing campaign performance, and refining personalized experiences, CNNs empower banks to make
precise, data-driven decisions that elevate customer satisfaction and loyalty. Comparative analyses demonstrate CNNs' superior
performance over traditional models like Random Forest and Logistic Regression, achieving accuracies up to 89% and F1 scores
of 88%, thereby highlighting their transformative potential in reshaping digital marketing strategies within the banking sector.
This research underscores the critical implications of adopting advanced deep learning techniques to meet the evolving
demands of customers in today's dynamic digital landscape.

Facial Expression Recognition (FER) is a crucial issue in human-machine interaction. It allows machines to act according to facial expression changes. However, acting in real time requires recognizing the expressions at video speed. Usually, the video speed differs from one device to another. However, one of the standard settings for shooting videos is 24 fps. This speed is considered as the low-end of what our brain can perceive as fluid video. From this perspective, to achieve a real-time FER the image analysis must be completed, strictly, in less than 0.042 second no matter how the background complexity is or how many faces exists in the scene. In this paper, a new pipeline has been proposed in order to recognize the fundamental facial expressions for more than one person in real world sequence videos. First, the pipeline takes as input a video and performs a face detection and tracking. Regions of Interest (ROI) are extracted from the detected face in order to extract the shape information when applying the Histogram of Oriented Gradient (HOG) descriptor. The number of features yield by HOG descriptor is reduced by means of a Linear Discriminant Analysis (LDA). Then, a deep data analysis was carried out, exploiting the pipeline, for the objective of setting up the LDA classifier. The analysis aimed at proving the suitability of the decision rule selected to separate the facial expression clusters in the LDA training phase. To conduct our analysis, we used ChonKanade(CK+)

Due to their high distinctiveness, robustness to illumination and simple computation, Histogram of Oriented Gradient (HOG) features have attracted much attention and achieved remarkable success in many computer vision tasks. In this paper, an innovative framework for driver drowsiness detection is proposed, where an adaptive descriptor that possesses the virtue of distinctiveness, robustness and compactness is formed from an improved version of HOG features based on binarized histograms of shifted orientations. The final HOG descriptor generated from binarized HOG features is fed to the trained Naïve Bayes (NB) classifier to make the final driver drowsiness determination. Experimental results on the publicly available NTHU-DDD dataset verify that the proposed framework has the potential to be a strong contender for several state-of-the-art baselines, by achieving a competitive detection accuracy of 85.62%, without loss of efficiency or stability.

Retrieving human actions from video databases is a paramount but challenging task in computer vision. In this work, we develop such a framework for robustly recognizing human actions in video sequences. The contribution of the paper is twofold. First a reliable neural model, the Multi-level Sigmoidal Neural Network (MSNN) as a classifier for the task of action recognition is presented. Second we unfold how the temporal shape variations can be accurately captured based on both temporal self-similarities and fuzzy log-polar histograms. When the method is evaluated on the popular KTH dataset, an average recognition rate of 94.3% is obtained. Such results have the potential to compare very favorably to those of other investigators published in the literature. Further the approach is amenable for real-time applications due to its low computational requirements.

Inspired by the overwhelming success of Histogram of Oriented Gradients (HOG) features in many vision tasks, in this paper, we present an innovative compact feature descriptor called fuzzy Histogram of Oriented Lines (f-HOL) for action recognition, which is a distinct variant of the HOG feature descriptor. The intuitive idea of these features is based on the observation that the slide area of the human body skeleton can be viewed as a spatiotemporal 3D surface, when observing a certain action being performed in a video. The f-HOL descriptor possesses an immense competitive advantage, not only of being quite robust to small geometric transformations where the small translation and rotations make no large fluctuations in histogram values, but also of not being very sensitive under varying illumination conditions. The extracted features are then fed into a discriminative conditional model based on Latent-Dynamic Conditional random fields (LDCRFs) to learn to recognize actions from video frames. When tested on the benchmark Weizmann dataset, the proposed framework substantially supersedes most existing state-of-the-art approaches, achieving an overall recognition rate of 98.2%. Furthermore, due to its low computational demands, the framework is properly amenable for integration into real-time applications.

This paper presents animated pose templates (APTs) for detecting short-term, long-term, and contextual actions from cluttered scenes in videos. Each pose template consists of two components: 1) a shape template with deformable parts represented in an And-node whose appearances are represented by the Histogram of Oriented Gradient (HOG) features, and 2) a motion template specifying the motion of the parts by the Histogram of Optical-Flows (HOF) features. A shape template may have more than one motion template represented by an Or-node. Therefore, each action is defined as a mixture (Or-node) of pose templates in an And-Or tree structure. While this pose template is suitable for detecting short-term action snippets in two to five frames, we extend it in two ways: 1) For long-term actions, we animate the pose templates by adding temporal constraints in a Hidden Markov Model (HMM), and 2) for contextual actions, we treat contextual objects as additional parts of the pose templates and add constraints that encode spatial correlations between parts. To train the model, we manually annotate part locations on several keyframes of each video and cluster them into pose templates using EM. This leaves the unknown parameters for our learning algorithm in two groups: 1) latent variables for the unannotated frames including pose-IDs and part locations, 2) model parameters shared by all training samples such as weights for HOG and HOF features, canonical part locations of each pose, coefficients penalizing pose-transition and part-deformation. To learn these parameters, we introduce a semi-supervised structural SVM algorithm that iterates between two steps: 1) learning (updating) model parameters using labeled data by solving a structural SVM optimization, and 2) imputing missing variables (i.e., detecting actions on unlabeled frames) with parameters learned from the previous step and progressively accepting high-score frames as newly labeled examples. This algorithm belongs to a family of optimization methods known as the Concave-Convex Procedure (CCCP) that converge to a local optimal solution. The inference algorithm consists of two components: 1) Detecting top candidates for the pose templates, and 2) computing the sequence of pose templates. Both are done by dynamic programming or, more precisely, beam search. In experiments, we demonstrate that this method is capable of discovering salient poses of actions as well as interactions with contextual objects. We test our method on several public action data sets and a challenging outdoor contextual action data set collected by ourselves. The results show that our model achieves comparable or better performance compared to state-of-the-art methods.

In order to solve the problem of pedestrian detection performance, the described operator was improved. In this paper, semantic local binary pattern (SLBP) and histogram of oriented gradient (HOG) are combined as new feature operator. This feature method would enrich the information and enhance the detection performance. And then histogram intersection kernel support vector machine (HIKSVM) classifier is trained by the augment feature. Because the time cost is too large by the conventional SVM. HIKSVM could make up this drawback, and significantly reduce the training time. The experiments on the INRIA pedestrian dataset show that the method obtained significant improvement in accuracy comparing to HOG descriptors.

This paper proposes a feature-based technique to detect pedestrians and recognize vehicles within thermal images that have been captured during nighttime. The proposed technique applies the support vector machine (SVM) classifier on CENsus Transformed histogRam Oriented Gradient (CENTROG) features in order to classify and detect humans and/or vehicles. Although thermal images suffer from low image resolution, lack of colour and poor texture information, they offer the advantage of being unaffected by high intensity light sources such as vehicle headlights which tend to render normal images unsuitable for nighttime image capturing and subsequent analysis. Since contour is the most distinctive feature within thermal images, CENTROG is used to capture this feature information and is used within the experiments. The experimental results so obtained were compared with those obtained by employing the CENsus TRansformed hISTogram (CENTRIST). Experimental results revealed that CENTROG offers better detection and classification accuracy for both pedestrian and detection and vehicle type recognition.

United States foreign policy has sought to maintain primacy on the global stage, but every U.S. president has maintained U.S. foreign policy in their own context. Counter to previous administrations, President Donald Trump has focused U.S. foreign policy through the lens of transactional realism concentrating on counterterrorism and challenges to U.S. interests and allies in the Middle East. Trump has criticized his predecessors for their policies of deep engagement leading to a deterioration of U.S. ‘Greatness,’ wasted tax dollars, and wasted American lives. Rooted in his background as an entrepreneur, President Trump has centered foreign policy around “America First.” However, U.S. interests in Syria are complicated by a multitude of actors involved in the Syrian conflict. Consequently, President Trump utilizes smart power with limited engagement in Syria based on cooperation  with regional and international actors, particularly with regards to the ‘War on Terror.’ This strategy f...

Thee fficient face recognitions ystems aret hose which area ble to achieveh igher recognitionr ate with lower computational cost. To develops uch systems bothf eaturerepresentationand classification methodshouldbeaccurate andl ess time consuming.Aiming to satisfyt hese criteria we coupled theH OG descriptor (Histograms of Oriented Gradients) with theRandomForest classifier (RF). Althoughrarelyusedinfacerecognition, HOGhaveprovent obeapower descriptor in this taskw ithalower computational time.A s regards classification method, recent works have shownthatapart from their accuracy when compared with its competitors, Random Forest exhibits al ow computational time in both traininga nd testing phase.E xperimental results on ORL databaseh ave demonstrated theefficiency of this combination.

Skin cancer is an exquisite disease globally nowadays. Because of the poor contrast and apparent resemblance between skin and lesions, automatic identification of skin cancer is complicated. The rate of human death can be massively reduced if melanoma skin cancer can be detected quickly using dermoscopy images. In this research, an anisotropic diffusion filtering method is used on dermoscopy images to remove multiplicative speckle noise and the fast-bounding box (FBB) method is applied to segment the skin cancer region. Furthermore, the paper consists of two feature extractor parts. One of the two features extractor parts is the hybrid feature extractor (HFE) part and another is the convolutional neural network VGG19 based CNN feature extractor part. The HFE portion combines three feature extraction approaches into a single fused feature vector: Histogram-Oriented Gradient (HOG), Local Binary Pattern (LBP), and Speed Up Robust Feature (SURF). The CNN method also is used to extract a...

Histogram of Oriented Gradients (HOG) based methods for the detection of humans have become one of the most reliable methods of detecting pedestrians with a single passive imaging camera. However, they are not 100 percent reliable. This paper presents an improved tracker for the monitoring of pedestrians within images. The Parallel HOG and Correlation Tracking (PHACT) algorithm utilises self learning to overcome the drifting problem. A detection algorithm that utilises HOG features runs in parallel to an adaptive and stateful correlator. The combination of both acting in a cascade provides a much more robust tracker than the two components separately could produce.

Tuberculosis (TB) is a severe infection that mostly affects the lungs and kills millions of people's lives every year. Tuberculosis can be diagnosed using chest X-rays (CXR) and data-driven deep learning (DL) approaches. Because of its better automated feature extraction capability, convolutional neural networks (CNNs) trained on natural images are particularly effective in image categorization. A combination of 3001 normal and 3001 TB CXR images was gathered for this study from different accessible public datasets. Ten different deep CNNs (Resnet50, Resnet101, Resnet152, InceptionV3, VGG16, VGG19, DenseNet121, DenseNet169, DenseNet201, MobileNet) are trained and tested for identifying TB and normal cases. This study presents a deep CNN approach based on histogram matched CXR images that does not require object segmentation of interest, and this coupled methodology of histogram matching with the CXRs improves the accuracy and detection performance of CNN models for TB detection. Furthermore, this research contains two separate experiments that used CXR images with and without histogram matching to classify TB and non-TB CXRs using deep CNNs. It was able to accurately detect TB from CXR images using pre-processing, data augmentation, and deep CNN models. Without histogram matching the best accuracy, sensitivity, specificity, precision and F1-score in the detection of TB using CXR images among ten models are 99.25%, 99.48%, 99.52%, 99.48% and 99.22% respectively. With histogram matching the best accuracy, sensitivity, specificity, precision and F1-score are 99.58%, 99.82%, 99.67%, 99.65% and 99.56% respectively. The proposed methodology, which has cutting-edge performance, will be useful in computer-assisted TB diagnosis and aids in minimizing irregularities in TB detection in developing countries.

Face detection is the main building block on which all automatic systems dealing with human faces is built. For example, a face recognition system must rely on face detection to process an input image and determine which areas contain human faces. These areas then become the input for the face recognition system for further processing. This paper presents a face detection system designed to detect frontal faces. The system uses Haar wavelet coefficients and local histogram correlation as differentiating features. Our proposed system is trained using 100 training images. Our experiments show that the proposed system performed well during testing, achieving a detection rate of 91.5%.