object detection

The work is devoted to the characteristic properties analysis of the sea surface image on video stream frames, which makes it possible to make a decision about the detection of object images on them in the absence of a priori information about objects, which is important in the development of navigation systems, security and surveillance systems. In this paper, it is proposes to analyze the values of the normalized cross-correlation coefficients of a given image fragment with higher dimension fragments on sequential video stream frames to identify the distinctive properties of the agitated sea surface and floating objects images. The computational experiments were carried out and have shown that the analysis of the results of calculating the frame fragments cross-correlations allows us to estimate the amount of displacement and distortion of a given image fragment. The results of computational experiments demonstrate the presence of differences in the values of the corresponding cross-correlation coefficients for the sea surface images with different agitation degrees, containing and not containing an image of the object. Based on the analysis of the proposed correlation properties of the sea surface images, a decisive rule for selecting fragments of frames containing an image of an object is formulated, the use of which, in many cases, allows to detection of fragments of object images correctly.

The process of controlling railway systems is one of the important and effective topics in the process of maintaining the flow of trains’ movement and organizing the travel process, as well as providing early readings of any defect or problem that occurs in the railway network to avoid, treat accidents and ensure a safe environment for the movement of train cars across the geographical area.  Therefore, a continuous follow-up of the railway condition must be provided to ensure that services continue to be provided. Intelligent railway maintenance improves safety and efficiency. This work presents the design and implementation of a real-time monitoring system for railways based on WSN. This study proposes a system consisting of a base station server and a Rail controller. The Base Station (BS) can automatically monitor and control most railway paths. For that, a classification-based deep learning model for object detection near the railway and making appropriate decisions were proposed. To improve classification-model performance, the yolov3 algorithm for object detection was proposed. On the Rail controller side, the Raspberry Pi 4 was utilized as a low-cost processing unit that can be used as a control unit to control some processes, such as streaming video from a camera, gathering information from railway sensors, and sending data to the central station server (PC) by using WiFi protocol. The model can detect and control railway issues in real time by receiving streaming data and directly detecting, classifying issues, and making the best decisions. By alarming or controlling the desired train and stopping it.

In recent years, development of the machine learning algorithms has led to the creation of intelligent surveillance systems. Thanks to the machine learning, it is possible to perform intelligent surveillance by recognizing people's facial features, classifying their age and gender, and detecting objects around instead of ordinary surveillance. In this study, a novel algorithm has been developed that classifies people's age and gender with a high accuracy rate. In addition, a novel object recognition algorithm has been developed that detects objects quickly and with high accuracy. In this study, age and gender classification was made based on the facial features of people using Convolutional Neural Network (CNN) architecture. Secondly, object detection was performed using different machine learning algorithms and the performance of the different machine learning algorithms was compared in terms of median average precision and inference time. The accuracy of the age and gender classification algorithm was tested using the Adience dataset and the results were graphed. The experimental results show that age and gender classification algorithms successfully classify people's age and gender. Then, the performances of object detection algorithms were tested using the COCO dataset and the results were presented in graphics. The experimental results stress that machine learning algorithms can successfully detect objects.

Object tracking and detection are among the most significant jobs in computer vision, having many applications in areas, which includes autonomous vehicle tracking, robotics, as well as traffic monitoring. Several studies have been conducted in past years. However, since detecting various problems, for instance, fast motion, illumination variations, as well as occlusion, study in this field persists. Furthermore, deep convolutional neural networks (DCNNs) have grown increasingly significant for object detection as deep learning (DL) techniques have advanced. As a result, numerous approaches for object detection are studied in this research, as well as a comprehensive. This project encompasses backbone networks, loss functions and training strategies, classical object detection architectures, complex problems, datasets and evaluation metrics, applications, future development directions, as well as a review and analysis of DL-based object detection techniques conducted in previous years. Experts in the field of object detection will benefit from this review article.

Many health systems face extraordinary problems because of the ongoing COVID-19 outbreak and new variations. Multiple regulatory authorities have made it mandatory to maintain a safe distance, particularly in public settings where large groups of people are likely to come into contact, such as sports arenas, public transportation, workplaces as well as shopping malls. Nevertheless, keeping a safe distance (two meters), adjusting multiple model detection errors or accuracy as well as deployment prerequisites, great number of people, facial expression, view angle, low-resolution images, detection model deployment on computers having restricted processing power, and the shortage of a real-world dataset have all made compliance and adherence to proper distancing social difficult. As a result, this survey examines and contrasts the most important past deep learning (DL)-based social distance research. Here, the survey presents a new fine-grained taxonomy that classifies the present state-of-the-art DL-based object detection for detecting distance in terms of several dimensions, such as detection, input data, evaluation methodologies, as well as testing, based on a thorough review. Each facet is then divided into categories based on a variety of factors. In addition, this survey analyses and evaluates the associated experimental techniques suggested as DL-based object detection. Finally, this survey examines DL's role in social distance, object detection datasets impact, as well as the proposed approaches efficacy by assessing the experimental research. The results show that more work is needed to enhance the existing state-of-the-art. Ultimately, open research difficulties are recognized, as well as prospective DL research areas are suggested for future research.

In today's world, where speech recognition has become an integral part of our daily lives, the need for systems equipped with this technology has increased dramatically in the past few years. This research aims to locate the two selected Persian words in any given audio file. For this purpose, two standard and native datasets were prepared for this model one for train and the other for the test. Both datasets were converted into images of audio waveforms. Using the object detection technique, the model could extract different bounding boxes for each test audio, and then each box image goes through a CNN classifier and returns a corresponding label. Finally, a threshold is set so that only boxes with high accuracy are displayed as output. The results showed 93% accuracy for the CNN classifier and 50% accuracy for testing the model with object detection.

Large parts of the world's forests are threatened by fires. These fires happen continuously every month around the globe. They are very costly to society and cause serious damage to the ecosystem. This raises the necessity to build a detection system to intervene early and take action. Fire and smoke have various colours, textures, and shapes, which are challenging to detect. In the modern world, neural networks are used extensively in most fields of human activities. For the detection of fire and smoke, we suggest a deep learning technology using transfer learning to extract features of forest fire and smoke. We used a pre-trained Inception-ResNet-v2 network on the ImageNet dataset to be trained on our dataset which consists of 1,102 images for each fire and smoke class. The classification accuracy, precision, recall, F1-Score, and specificity were 99.09\%, 100\%, 98.08\%, 99.09\%, and 98.30\%, respectively. This model has been deployed on a Raspberry Pi device with a camera. For real-time detection, we used the Open CV library to read the camera stream frame by frame and predict the probability of fire or smoke.

Corona virus sickness has become a big public health issue in 2019. Because of its contact-transparent characteristics, it is rapidly spreading. The use of a face mask is among the most efficient methods for preventing the transmission of the Covid-19 virus. Wearing the face mask alone can cut the chance of catching the virus by over 70\%. Consequently, World Health Organization (WHO) advised wearing masks in crowded places as precautionary measures. Because of the incorrect use of facial masks, illnesses have spread rapidly in some locations. To solve this challenge, we needed a reliable mask monitoring system. Numerous government entities are attempting to make wearing a face mask mandatory; this process can be facilitated by using face mask detection software based on AI and image processing techniques. For face detection, helmet detection, and mask detection, the approaches mentioned in the article utilize Machine learning, Deep learning, and many other approaches. It will be simple to distinguish between persons having masks and those who are not having masks using all of these ways. The effectiveness of mask detectors must be improved immediately. In this article, we will explain the techniques for face mask detection with a literature review and drawbacks for each technique.