Closed-circuit Television Images Research Articles

Evaluating Methods for Violence Classification and Firearm Detection in Indoor CCTV Environment

The adoption of security systems based on computer vision for violence detection has the potential to significantly improve safety in various public and private properties. However, developing these systems can be extremely challenging.We can choose to use classification models to identify violence in images or also use object detection models to identify firearms, which may indicate robbery. Additionally, when developing such systems focused on private environments, we encounter specific challenges, such as obtaining appropriate datasets to train the algorithms. Many publicly available datasets for violence detection consist of outdoor images, with elements such as streets and cars, which do not adequately reflect the nuances and unique characteristics of private properties. In this work, we evaluate both learned and handcrafted features to classify videos as 'violence' or 'non-violence' across a variety of datasets, including a new dataset composed exclusively of closed-circuit television (CCTV) images. Additionally, we propose a new dataset for firearm detection in CCTV images and conduct some experiments using YoloV8. In this way, we hope to provide a clearer insight into the possible decisions when developing a security system for indoor environments.

Evaluation of Preferences for a Thermal-Camera-Based Abnormal Situation Detection Service via the Integrated Fuzzy AHP/TOPSIS Model

Research related to thermal cameras, which are major control measures, is increasing to overcome the limitations of closed-circuit television (CCTV) images. Thermal cameras have the advantage of easily detecting objects at night and of being able to identify initial signs of dangerous situations owing to changes in temperature. However, research on thermal cameras from a comprehensive perspective for practical urban control is insufficient. Accordingly, this study presents a thermal camera-based abnormal-situation detection service that can supplement/replace CCTV image analysis and evaluate service preferences. We suggested an integrated Fuzzy AHP/TOPSIS model, which induces a more reasonable selection to support the decision-making of the demand for introducing thermography cameras. We found that developers highly evaluated services that can identify early signs of dangerous situations by detecting temperature changes in heat, which is the core principle of thermography cameras (e.g., pre-fire phenomenon), while local governments highly evaluated control services related to citizen safety (e.g., pedestrian detection at night). Clearly, while selecting an effective service model, the opinions of experts with a high understanding of the technology itself and operators who actually manage ser-vices should be appropriately reflected. This study contributes to the literature and provides the basic foundation for the development of services utilizing thermography cameras by presenting a thermography camera-based abnormal situation detection service and selection methods and joint decision-making engagement between developers and operators.

The Performance of K-Nearest Neighbor (KNN) Approach for Estimating Extreme Rain Events based on CCTV Images Camera Data

Abstract—Hydrometeorological disasters such as flooding in urban areas are a big problem that must be managed. Indonesia as part of the maritime continent, has high rainfall variability, both temporally and spatially. Unfortunately, the density of instruments for measuring rainfall is still low. To solve the problem, this research will try to utilize and modify Closed Circuit Television (CCTV) cameras which have a large number in terms of quantity as instruments for measuring rainfall. The purpose of this research is to obtain rainfall image information and data generated by CCTV cameras. The image data is converted to quantitative rainfall data. The method used is the K-NN algorithm and machine learning. The research location is located in a corner of the city of Bandung with a geographical position of 60 53”30.49'S and 107.035” 12.27' E. The results of this research show that the K-NN algorithm can be applied to estimate rainfall data from CCTV images with an accuracy of more than 98%. The level of accuracy generated between CCTV camera image data and AWS is 94%. The level of accuracy is high means that CCTV camera image data can represent or be converted into quantitative rainfall data.

Deep residual neural networks for increasing the resolution of CCTV images

Images from closed-circuit television (CCTV) cameras are often stored with poor resolution due to technical and data storage limitations, reducing the utility of these images. In this study, we propose that informed design and training of an enhanced deep residual network (EDSR) can increase the resolution of CCTV images. This work is the first published application of the EDSR methodology for increasing CCTV image resolution. We utilized a dataset of 986 CCTV images that trained and tested an EDSR for generating high-resolution images from a low-resolution dataset. Implementing an EDSR model, we demonstrate that with 60 layers, performing image super-resolution is extremely effective, producing a peak signal-to-noise ratio value of 76.83 decibels. Furthermore, this technique may be used more generally as a technique for file compression enabling the storage of images at smaller file sizes at low-resolution and the ability to reproduce the same images in high-resolution form.

Spatio-Temporal Network for Sea Fog Forecasting

Sea fog can seriously affect schedules and safety by reducing visibility during marine transportation. Therefore, the forecasting of sea fog is an important issue in preventing accidents. Recently, in order to forecast sea fog, several deep learning methods have been applied to time series data consisting of meteorological and oceanographic observations or image data to predict fog. However, these methods only use a single image without considering meteorological and temporal characteristics. In this study, we propose a multi-modal learning method to improve the forecasting accuracy of sea fog using convolutional neural network (CNN) and gated recurrent unit (GRU) models. CNN and GRU extract useful features from closed-circuit television (CCTV) images and multivariate time series data, respectively. CCTV images and time series data collected at Daesan Port in South Korea from 1 March 2018 to 14 February 2021 by Korea Hydrographic and Oceanographic Agency (KHOA) were used to evaluate the proposed method. We compare the proposed method with deep learning methods that only consider temporal information or spatial information. The results indicate that the proposed method using both temporal and spatial information at the same time shows superior accuracy.

Gun Detection in CCTV Images using HAAR-Like Features

Automated video-based surveillance is an important area of research to assist the security personnel to detect the incident of any abnormal events in the surroundings. The objective of this paper is to develop a framework for automatic gun detection using closed-circuit television (CCTV) images. The methodology presented in this paper involves the development of a framework for automatic gun detection using closed-circuit television (CCTV) images, with the aim of enhancing the surveillance of crime and improving human security. The proposed approach consists of a dataset of CCTV images containing instances of guns, as well as non-gun images for comparison. These images would be used to train the proposed algorithm to recognize and identify guns in future CCTV images. The proposed framework is designed for an indoor environment and uses Haar-like features for gun detection. The proposed system involves the installation of CCTV cameras in a suitable corner of an indoor environment for surveillance. The CCTV cameras capture the scene and the frames of the scene are compared with a predefined dataset for automatic gun detection. The proposed approach draws a bounding box and raises an alarm if it detects a gun in a frame extracted from a captured scene. This provides a visual indication of the presence of a gun, making it easier for relevant authorities to quickly identify and respond to the threat. The proposed system shows promising results in real-time applications and about 90% accuracy has been achieved.

ACF: An Armed CCTV Footage Dataset for Enhancing Weapon Detection.

Thailand, like other countries worldwide, has experienced instability in recent years. If current trends continue, the number of crimes endangering people or property will expand. Closed-circuit television (CCTV) technology is now commonly utilized for surveillance and monitoring to ensure people’s safety. A weapon detection system can help police officers with limited staff minimize their workload through on-screen surveillance. Since CCTV footage captures the entire incident scenario, weapon detection becomes challenging due to the small weapon objects in the footage. Due to public datasets providing inadequate information on our interested scope of CCTV image’s weapon detection, an Armed CCTV Footage (ACF) dataset, the self-collected mockup CCTV footage of pedestrians armed with pistols and knives, was collected for different scenarios. This study aimed to present an image tilling-based deep learning for small weapon object detection. The experiments were conducted on a public benchmark dataset (Mock Attack) to evaluate the detection performance. The proposed tilling approach achieved a significantly better mAP of 10.22 times. The image tiling approach was used to train different object detection models to analyze the improvement. On SSD MobileNet V2, the tiling ACF Dataset achieved an mAP of 0.758 on the pistol and knife evaluation. The proposed method for enhancing small weapon detection by using the tiling approach with our ACF Dataset can significantly enhance the performance of weapon detection.

A Real-Time Framework for Human Face Detection and Recognition in CCTV Images

This paper aims to develop a machine learning and deep learning-based real-time framework for detecting and recognizing human faces in closed-circuit television (CCTV) images. The traditional CCTV system needs a human for 24/7 monitoring, which is costly and insufficient. The automatic recognition system of faces in CCTV images with minimum human intervention and reduced cost can help many organizations, such as law enforcement, identifying the suspects, missing people, and people entering a restricted territory. However, image-based recognition has many issues, such as scaling, rotation, cluttered backgrounds, and variation in light intensity. This paper aims to develop a CCTV image-based human face recognition system using different techniques for feature extraction and face recognition. The proposed system includes image acquisition from CCTV, image preprocessing, face detection, localization, extraction from the acquired images, and recognition. We use two feature extraction algorithms, principal component analysis (PCA) and convolutional neural network (CNN). We use and compare the performance of the algorithms K-nearest neighbor (KNN), decision tree, random forest, and CNN. The recognition is done by applying these techniques to the dataset with more than 40K acquired real-time images at different settings such as light level, rotation, and scaling for simulation and performance evaluation. Finally, we recognized faces with a minimum computing time and an accuracy of more than 90%.

Automatic Human Detection Using Reinforced Faster-RCNN for Electricity Conservation System

Electricity conservation systems are designed to conserve electricity to manage the bridge between the high raising demand and the production. Such systems have been so far using sensors to detect the necessity which adds an additional cost to the setup. Closed-circuit Television (CCTV) has been installed in almost everywhere around us especially in commercial places. Interpretation of these CCTV images is being carried out for various reasons to elicit the information from it. Hence a framework for electricity conservation that enables the electricity supply only when required, using existing resources would be a cost effective conservation system. Such a framework using a deep learning model based on Faster-RCNN is developed, which makes use of these CCTV images to detect the presence or absence of a human in a place. An Arduino micro-controller is embedded to this framework which automatically turns on/off the electricity based on human's presence/absence respectively. The proposed approach is demonstrated on CHOKE POINT dataset and two real time datasets which images from CCTV footages. F-measure, Accuracy scores (AUC score) and training time are the metrics for which the model is evaluated. An average accuracy rate of 82% is obtained by hyper-parameter tuning and using Adam optimization technique. This lays the underpinning for designing automatic frameworks for electricity conservation systems using existing resources.

Particulate Matter Estimation from Public Weather Data and Closed-Circuit Television Images

This article proposes a method of estimating the concentrations of particulate matter (PM2.5 and PM10) using public data, including road-traffic closed-circuit television (CCTV) images, Smart Seoul City data sensor environment information, and Korea Meteorological Administrationdata. The region-of-interest images and full scenes derived from CCTV footage were used as the basis for the deep learning model, which combines a convolutional neural network and long short-term memory, to establish the particulate matter (PM) concentration prediction methodology. In the experiment, the prediction accuracies corresponding to various types of mean values were calculated by training the model with various mean measurement values for the surface PM2.5 and PM10 concentrations, as well as the corresponding CCTV images and weather data at different time points. In the experiments performed under relatively stable PM concentrations, R2 generally exceeded 0.9 and tended to increase with an increasing range of mean concentration values. In particular, in sections with rapid changes in the PM concentration within an hourly interval, higher R2 values were obtained by the model trained with the average PM concentrations of the time series before and after image capture, outperforming the method that used prior mean observation values and better reflecting the current PM concentration.

Detection of Wave Parameters Using CCTV Images-Based on Deep Learning Algorithm

Yoon, J.; Kim, S.; Kim, I., and Song, D., 2021. Detection of wave parameters using CCTV images-based on deep learning algorithm. In: Lee, J.L.; Suh, K.-S.; Lee, B.; Shin, S., Lee, J. (eds.), Crisis and Integrated Management for Coastal and Marine Safety. Journal of Coastal Research, Special Issue No. 114, pp. 281–284. Coconut Creek (Florida), ISSN 0749-0208. The closed-circuit television (CCTV) systems have been used for various applications in oceanic research fields. CCTV is an effective installation method for offshore monitoring that can conveniently and continuously observe the coastal region using the visible or infrared light in the electromagnetic spectrum. However, in the case of wave-observing CCTV images, there is a problem: it is difficult to extract the corresponding feature points between the video frames, which causes of inaccurate measurement of wave parameters. Recently, in the field of machine learning technology, deep learning-based video analysis that does not require the detection of feature points has been proposed. In this paper, we analyze wave-observing CCTV images with deep learning method in particular convolution neural network, and find the optimal method of observing waves through CCTV by three different datasets. For training, the CCTV data from the Samcheok Beach in the Republic of Korea were used. Wave observation data, which were used as input for machine learning, were collected by an acoustic wave and current profiler (AWAC). Moreover, to make up for insufficient observation data, the data to enhance the prediction model through deep learning were augmented. The results of this study confirm that the deep learning algorithm can be utilized to detect wave period and wave height. Consequently, continuous CCTV images can be combined with deep learning algorithm to clarify the complex characteristics of wave distribution and its temporal changes.

Method for Obtaining Better Traffic Survey Data

Road traffic surveys determine the number and type of vehicles passing by a specific point over a certain period of time. The manual estimation of the number and type of vehicles from images captured by a camera is the most commonly used method. However, this method has the disadvantage of requiring high amounts of manpower and cost. Recently, methods of automating traffic volume surveys using sensors or deep learning have been widely attempted, but there is the disadvantage that a person must finally manually verify the data in order to ensure that they are reliable. In order to address these shortcomings, we propose a method for efficiently conducting road traffic volume surveys and obtaining highly reliable data. The proposed method detects vehicles on the road from CCTV (Closed-circuit television) images and classifies vehicle types using deep learning or a similar method. After that, it automatically informs the user of candidates with a high probability of error and provides a method for efficient verification. The performance of the proposed method was tested using a data set collected by an actual road traffic survey company. As a result, we proved that our method shows better accuracy than the previous method. The proposed method can reduce the labor and cost in road traffic volume surveys, and increase the reliability of the data due to more accurate results.

Towards an automated condition assessment framework of underground sewer pipes based on closed-circuit television (CCTV) images

There is a growing trend of using computer vision techniques for interpreting Closed-Circuit Television (CCTV) inspection videos of sewer pipes. Previous studies mainly focus on detecting defect types and locations in CCTV images, yet limited systematic approaches are available for automatically evaluating defect severity and sewer condition with references to existing standards. This study proposes a framework for evaluating defect severity and sewer condition automatically from CCTV images using computer vision methods, which includes (1) required information definition for sewer condition assessment, (2) pipe joint detection and fitting by image processing techniques to obtain cross section area, (3) defect detection and segmentation to obtain defect type, location and area, and (4) evaluation of defect severity and sewer condition. Particularly, three deep learning -based defect detection models are developed, among which the model based on Faster R-CNN (regional convolution neural network) outperforms others with higher accuracy and is used for detecting defect type and location in the image. Meanwhile, an innovative semantic segmentation model is applied for segmenting defects to extract defect area in the image. In the validation, our framework performs well in defect detection with an average precision, recall and F1 of 88.99%, 87.96%, and 88.21% respectively. More importantly, the framework evaluates Operation and Maintenance (O&M) defects more accurately by precise calculation and generates the overall condition gradings that are mostly consistent with professional inspectors, only with an average deviation of 3.06%. Our framework can assist the review of inspection videos and lays the basis for fully automated sewer assessment and maintenance planning in the future. Without constraints on the assessment codes and computer vision methods, the framework is adaptable to evaluating sewer condition in different regions and can achieve better performance by integrating with cutting-edge vision techniques.

RHC: A Dataset for In-Room and Out-Room Human Counting

Abstract In recent years, research in human counting from CCTV (Closed Circuit Television) images have found an increasing demand to be deployed in real-world applications. The applications have been implemented in various settings, both indoor and outdoor. In the case of indoor setting, we found a type of room setting that conveys a problem to human counting model if we need to count only humans inside a room. With this respect, we present RHC (Room Human Counting) dataset, which images are captured in the aforementioned setting. The dataset can be used to develop a robust model that can differentiate between humans inside and outside a room. The dataset is publicly available at https://data.mendeley.com/datasets/vt5c8h6kmh/1.

Rain detection in image using convolutional neural network

Weather is a phenomenon occurs in the earth’s atmosphere. Weather affects human daily activities, especially outdoor activities. Weather observations including rainfall observation in Indonesia conducted by Meteorology, Climatology, and Geophysical Agency (BMKG). BMKG facing a major problem in terms of rainfall data spatial density. The insufficient amount and unevenly distributed rainfall measurement instrument, are two main factors contributing to rainfall data special density problems. One of the very prominent methods to gain a larger amount of rainfall measurement location is using the image obtained from existing Closed Circuit Television (CCTV) spread over vast areas, especially in the Jakarta region. The approach to recognize and classify the rainfall in a certain area from the CCTV image used in this research is the Convolutional Neural Network (CNN) method. The image data was taken from CCTV located in Kamal, Kalideres, West Jakarta. The images taken is split into two categories, the one that shows a rainy day and the one that shows a clear day. These two categories of images will be used as sample data to train CNN, an effort to obtain a suitable model. By using the CNN method, it’s possible to recognize and classify the rainfall condition within an image based on the model. Python is an open-source programming language that widely used nowadays to run CNN. The image classification using this CNN, scored approximately 98.30% of accuracy, which means that the model is optimal to recognize and classify rainfall conditions in a certain area based on the CCTV images.

Implementation of Image Enhancement Algorithm for Image Forensics using Mathlab

The purpose of this journal is to explain the implementation of the image enhancement algorithm for image forensics. Image Forensic deals with the types of digital evidence in the form of digital image files. One of the most commonly used digital devices in providing digital evidence for forensic analysis is CCTV (Closed-Circuit Television). CCTV images have a low quality such as noise, blur, lack of light intensity, etc., so that the image must be enhanced so that forensic analysis can be done. To enhance image quality, an application is needed by applying the image enhancement algorithm. The algorithm applied to the application is a Low Pass Filter to increase low pixel intensity, High Pass Filter to increase high pixel intensity, Median Filter to replaces the original pixel value with the pixel center value of the image, Mean Filter to replaces the original pixel value with a value the average pixel of the image, the Gaussian Filter for reducing noise in the image, the Wiener Filter to reduce blur in the image, the Histogram Equalization spreads the image histogram value, Contrast Stretching to stretch the contrast intensity in the image and Bicubic Interpolation to increase the image size and resize the image. In this study, the application was built using MATLAB and the testing process for each algorithm was based on Timing-Run, MSE and PSNR parameters. From the test, the average MSE value is 1058.512083 and the PSNR value is 541.61875 dB, which means that the resulting image has a fairly high level of similarity and the average time needed to process the algorithm for the image is 0.114627915 seconds.

Evidence of Long-Distance Droplet Transmission of SARS-CoV-2 by Direct Air Flow in a Restaurant in Korea.

BackgroundThe transmission mode of severe acute respiratory syndrome coronavirus 2 is primarily known as droplet transmission. However, a recent argument has emerged about the possibility of airborne transmission. On June 17, there was a coronavirus disease 2019 (COVID-19) outbreak in Korea associated with long distance droplet transmission.MethodsThe epidemiological investigation was implemented based on personal interviews and data collection on closed-circuit television images, and cell phone location data. The epidemic investigation support system developed by the Korea Disease Control and Prevention Agency was used for contact tracing. At the restaurant considered the site of exposure, air flow direction and velocity, distances between cases, and movement of visitors were investigated.ResultsA total of 3 cases were identified in this outbreak, and maximum air flow velocity of 1.2 m/s was measured between the infector and infectee in a restaurant equipped with ceiling-type air conditioners. The index case was infected at a 6.5 m away from the infector and 5 minutes exposure without any direct or indirect contact.ConclusionDroplet transmission can occur at a distance greater than 2 m if there is direct air flow from an infected person. Therefore, updated guidelines involving prevention, contact tracing, and quarantine for COVID-19 are required for control of this highly contagious disease.

CNN-Based Person Detection Using Infrared Images for Night-Time Intrusion Warning Systems.

Night-time surveillance is important for safety and security purposes. For this reason, several studies have attempted to automatically detect people intruding into restricted areas by using infrared cameras. However, detecting people from infrared CCTV (closed-circuit television) is challenging because they are usually installed in overhead locations and people only occupy small regions in the resulting image. Therefore, this study proposes an accurate and efficient method for detecting people in infrared CCTV images during the night-time. For this purpose, three different infrared image datasets were constructed; two obtained from an infrared CCTV installed on a public beach and another obtained from a forward looking infrared (FLIR) camera installed on a pedestrian bridge. Moreover, a convolution neural network (CNN)-based pixel-wise classifier for fine-grained person detection was implemented. The detection performance of the proposed method was compared against five conventional detection methods. The results demonstrate that the proposed CNN-based human detection approach outperforms conventional detection approaches in all datasets. Especially, the proposed method maintained F1 scores of above 80% in object-level detection for all datasets. By improving the performance of human detection from infrared images, we expect that this research will contribute to the safety and security of public areas during night-time.

Direct generation of level of service maps from images using convolutional and long short-term memory networks

Congestion in transport stations could result in stampede development and deadly crush situations. Closed circuit television (CCTV) cameras enable station managers to monitor the crowd and reduce overcrowding risks. However, identifying congestion conditions is a very laborious task for a human operator who has to monitor multiple locations at the same time. This paper presents a new approach to automated image-based identification of congestion as measured by level of service (LOS), which is the most widely accepted standard for measuring congestion. Existing methods for measuring LOS based on crowd density estimation from images have the disadvantages that, crowd density cannot be estimated accurately. In addition, the calculation of flow parameters involves a complex process, and consequently these parameters are not indicative of congestion in real-time. This paper proposes a novel method based on machine learning to directly classify LOS without calculating flow parameters. In the proposed method, visual features extracted by a deep convolutional neural network are classified using a support vector machine classifier and the classification results are further refined by using a long short-term memory network. A second contribution of this research is to develop a web-based LOS map visualization platform to monitor pedestrian distribution and variation of distribution in real-time. Experimental evaluation at Flinders Street Station in Melbourne shows that this method can achieve an accuracy of 81.9% and efficiency of 0.40 seconds per frame in LOS classification using CCTV images.

Automated detection of sewer pipe defects in closed-circuit television images using deep learning techniques

Abstract Sanitary sewer systems are designed to collect and transport sanitary wastewater and stormwater. Pipe inspection is important in identifying both the type and location of pipe defects to maintain the normal sewer operations. Closed-circuit television (CCTV) has been commonly utilized for sewer pipe inspection. Currently, interpretation of the CCTV images is mostly conducted manually to identify the defect type and location, which is time-consuming, labor-intensive and inaccurate. Conventional computer vision techniques are explored for automated interpretation of CCTV images, but such process requires large amount of image pre-processing and the design of complex feature extractor for certain cases. In this study, an automated approach is developed for detecting sewer pipe defects based on a deep learning technique namely faster region-based convolutional neural network (faster R-CNN). The detection model is trained using 3000 images collected from CCTV inspection videos of sewer pipes. After training, the model is evaluated in terms of detection accuracy and computation cost using mean average precision (mAP), missing rate, detection speed and training time. The proposed approach is demonstrated to be applicable for detecting sewer pipe defects accurately with high accuracy and fast speed. In addition, a new model is constructed and several hyper-parameters are adjusted to study the influential factors of the proposed approach. The experiment results demonstrate that dataset size, initialization network type and training mode, and network hyper-parameters have influence on model performance. Specifically, the increase of dataset size and convolutional layers can improve the model accuracy. The adjustment of hyper-parameters such as filter dimensions or stride values contributes to higher detection accuracy, achieving an mAP of 83%. The study lays the foundation for applying deep learning techniques in sewer pipe defect detection as well as addressing similar issues for construction and facility management.