Closed-circuit Television Research Articles

INTELLIGENT VIDEO SURVEILLANCE USING DEEP LEARNING

Video surveillance, also known as CCTV (closed-circuit television), is a rapidly growing industry that has been around for more than 30 years and has seen its fair share of technological advancements. In today's world, video surveillance has become an essential component for ensuring public safety. Security can be defined in a variety of ways depending on the context, such as theft detection, violence detection, explosion risk, and so on. The term "security" in crowded public places refers to almost any type of abnormal event.Intelligent video surveillance delivers cutting-edge smart security by recording unexpected activities in homes, offices, and public locations based on the preferences of the user. In the event of an abnormal incident, the video surveillance system will actively respond to detect actions in advance through real-time monitoring and promptly communicate data. The primary focus is on the use of deep learning techniques to provide tracking of a moving target, high-definition picture quality, and night vision technology triggered by motion sensors, which means the system isn't running when nothing is happening at your location, automatic audio and visual detection, video recording initiation, and detection of suspicious activities that can trigger and alert the systems in all climate conditions.Deep learning technology will be used in the data processing model design to visualise data for abnormal activities, and this design also proposes an intelligent surveillance system to quickly and effectively detect activities by sending a video image and an alert message to the web via real-time processing. Recent improvements in computer vision, particularly deep learning approaches, have opened up new possibilities for these systems, enhancing their capabilities and launching new research areas in this field.

HUMAN ABNORMAL ACTIVITY RECOGNITION

This paper brings forward one amongst the foremost significant applications of human suspicious activity recognition that is termed as anomaly detection. A key concern of any society today is providing safety to an individual. The main reason behind this concern is due to the constantly increasing activities causing threats, starting from deliberate ferocity to an injury caused through an accident. Simple installation of a traditional closed circuit television(CCTV) is not sufficient as it requires a person to continuously stay alert and monitor the cameras, which is quite inefficient. This call for the requirement to develop an security system which is fully automated system that recognizes anomalous activities in real time and brings instant help to the victims. Hence we proposed a system which will examine and detect the suspicious human action from real-time CCTV footage with help of machine learning techniques and generates the alert if the abnormal activity is occurred. The method is implemented on the dataset containing both normal and anomaly activity and experiment has shown better results.

Sistem Informasi Geografis (SIG) Titik CCTV Di Kota Jambi Berbasis Web

Requests for information regarding Closed Circuit Television (CCTV) in public areas by the people of Jambi City through the complaint channel, are increasingly being received by the Jambi City Communications and Information Office. The installation of Closed Circuit Television (CCTV) in public areas in Jambi City is one of the service innovations for the community in order to strengthen law enforcement, public order and public convenience and at the same time realize the Jambi City Government's Smart City program. CCTV installations in Jambi City have been carried out in various areas of Jambi City such as in various parking areas in government offices, at crossroads, and red lights. Even private business actors also support the installation of CCTV in their respective business areas. Advances in GIS technology can be applied to mapping the distribution of CCTV points in the city of Jambi to help collect information on the distribution of CCTV points in various public areas of Jambi City electronically so that they can improve excellent service for the community in the field of security and public comfort both for business actors and for visitors who came to the city of Jambi. So the final result of this research is the proposed design of a Web-Based Geographic Information System (GIS) application for CCTV Point in Jambi City. The system development method used is Prototype using Use Case Diagram modeling tools, Activity Diagrams, and Class Diagrams. This research is expected to provide complete information about the distribution of CCTV points in the Jambi City area, and become the initial foundation for integrating all CCTV management in Jambi City.

Optimal selective floor cleaning using deep learning algorithms and reconfigurable robot hTetro

Floor cleaning robots are widely used in public places like food courts, hospitals, and malls to perform frequent cleaning tasks. However, frequent cleaning tasks adversely impact the robot’s performance and utilize more cleaning accessories (such as brush, scrubber, and mopping pad). This work proposes a novel selective area cleaning/spot cleaning framework for indoor floor cleaning robots using RGB-D vision sensor-based Closed Circuit Television (CCTV) network, deep learning algorithms, and an optimal complete waypoints path planning method. In this scheme, the robot will clean only dirty areas instead of the whole region. The selective area cleaning/spot cleaning region is identified based on the combination of two strategies: tracing the human traffic patterns and detecting stains and trash on the floor. Here, a deep Simple Online and Real-time Tracking (SORT) human tracking algorithm was used to trace the high human traffic region and Single Shot Detector (SSD) MobileNet object detection framework for detecting the dirty region. Further, optimal shortest waypoint coverage path planning using evolutionary-based optimization was incorporated to traverse the robot efficiently to the designated selective area cleaning/spot cleaning regions. The experimental results show that the SSD MobileNet algorithm scored 90% accuracy for stain and trash detection on the floor. Further, compared to conventional methods, the evolutionary-based optimization path planning scheme reduces 15% percent of navigation time and 10% percent of energy consumption.

Object recognition in optical camera communication enabled by image restoration.

As an important branch of visible light communication (VLC), optical camera communication (OCC) has received increasing attention recently, owing to its availability and low cost of deployment by re-using cameras as VLC receivers. However, cameras on popular smartphones and/or closed-circuit television systems have their primary function for taking pictures and recognizing objects, where the recorded images with objects are inevitable to be distorted by the coded light under OCC. To this end, we propose and experimentally demonstrate an improved OCC system which is able to achieve data communication and object recognition simultaneously. Basically, we devise an image restoration (IR) scheme to repair the pixels damaged by modulated light during data transmission, and it hence provides better image input to realize object recognition. Moreover, to maintain a reasonable data rate of OCC, we also engineer an object avoidance (OA) scheme to remove the negative effect caused by the object background in OCC frame. Finally, we implement a prototype of the proposed system to verify its performance on object recognition and communication, and experimental results show that the proposed IR can bring an improvement over 37% in terms of object recognition accuracy comparing to the baseline under a data rate of 5 kbps.

Counting people inside a region-of-interest in CCTV footage with deep learning.

In recent years, the performance of people-counting models has been dramatically increased that they can be implemented in practical cases. However, the current models can only count all of the people captured in the inputted closed circuit television (CCTV) footage. Oftentimes, we only want to count people in a specific Region-of-Interest (RoI) in the footage. Unfortunately, simple approaches such as covering the area outside of the RoI are not applicable without degrading the performance of the models. Therefore, we developed a novel learning strategy that enables a deep-learning-based people counting model to count people only in a certain RoI. In the proposed method, the people counting model has two heads that are attached on top of a crowd counting backbone network. These two heads respectively learn to count people inside the RoI and negate the people count outside the RoI. We named this proposed method Gap Regularizer and tested it on ResNet-50, ResNet-101, CSRNet, and SFCN. The experiment results showed that Gap Regularizer can reduce the mean absolute error (MAE), root mean square error (RMSE), and grid average mean error (GAME) of ResNet-50, which is the smallest CNN model, with the highest reduction of 45.2%, 41.25%, and 46.43%, respectively. On shallow models such as the CSRNet, the regularizer can also drastically increase the SSIM by up to 248.65% in addition to reducing the MAE, RMSE, and GAME. The Gap Regularizer can also improve the performance of SFCN which is a deep CNN model with back-end features by up to 17.22% and 10.54% compared to its standard version. Moreover, the impacts of the Gap Regularizer on these two models are also generally statistically significant (P-value < 0.05) on the MOT17-09, MOT20-02, and RHC datasets. However, it has a limitation in which it is unable to make significant impacts on deep models without back-end features such as the ResNet-101.

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.

REGULATION OF THE USE OF CCTV AS A CRIME PREVENTION TECHNIQUE

Having been used for a number of years as part of the security apparatus of banks and stores, closed circuit television (CCTV) is increasingly becoming more prevalent on the streets of South Africa. CCTV monitoring takes place in a number of cities, including Johannesburg, Durban, Cape Town, Port Elizabeth and Pietermaritzburg. (My own interest in this matter derives from my involvement in the Justice Monitoring Project (JUMP), the brainchild of Prof Michael Cowling, which tasked students in 2004 to inter alia monitor theuse of CCTV evidence obtained from the camera network which is run as a joint venture of the South African Police Service (SAPS) and Business Against Crime in Pietermaritzburg.) Whilst it has been argued that CCTV cameras can make a meaningful contribution to enhancing the safety of our city streets, the question arises: at what cost? This paper seeks to examine this question in the context of a discussion of the use of CCTV as a law enforcement technique. The focus of the note will be on data collection rather than data collation, and the issue of admissibility of video evidence will not be dealt with. CCTV may be briefly described as a television system wherein signals are not publicly distributed and images are not broadcast. Instead, such images are transmitted from cameras to particular monitors serving a limited area. In the context of cameras located within a city, the images transmitted are primarily of pedestrians walking or transacting on the streets. As Froomkin has noted, whilst moving about in public is not truly anonymous, one at least enjoys the idea, largely consonant with reality in a city environment, of being able to move about with anonymity. This freedom is naturally threatened by privacy-destroying technology such as CCTV. It is evident that the use of such technology involves trying to balance the competing values of safety and privacy. As more cities make use of video surveillance, it appears as if privacy is on the losing side. Whilst this trend is understandable given the standard contradistinction between the tangible harms inherent in security concerns as opposed to the intangible harms occasioned by infringement on privacy, whether it is justifiable is another matter altogether. Spencer has raised a number of concerns about the tangible/intangible division as a framework for decision-making, namely (i) that the framework is incomplete and locks into short-term benefits and consequences rather than taking account of the long-term effects on privacy; (ii) that by embedding the decision in a concrete factual context the decision will inevitably be weighted in favour of preventing tangible harms; and (iii) that the tangible/intangible dichotomy is not as clear-cut as it may seem, since security proposals often serve intangible goals such as allaying fear, whilst privacy intrusions can have tangible consequences such as disrupting or inhibiting behaviour.

Does wearing a surgical mask influence face touching by healthcare workers? A retrospective cohort study.

At the start of the COVID-19 pandemic, healthcare workers (HCW) in our ED were advised against and actively discouraged from wearing masks when not seeing respiratory patients, as mask wearing was thought to increase the risk of droplet transmission by face touching. The primary objective of the present study was to determine whether HCW using face masks were more or less likely to touch their faces than those not wearing masks. We analysed six randomly selected hours of closed circuit television footage from our staff base. Face touches were recorded electronically by trained researchers. Generalised linear mixed models were used to compare the frequency and duration of face touches with and without face masks, controlling for individual clusters, adjusting for time of footage, duration on screen and staff role. Data were collected from 187 HCW. Masks were worn in 231 (36%) of 642 screen sessions. Wearing a mask did not significantly change the odds of face touching (odds ratio 0.55, 95% confidence interval [CI] 0.30-1.01, P=0.055) or duration of face touch (mean difference -1.45 s, 95% CI -8.84, 5.99, P=0.71). For mucosal areas, a significant reduction in the odds of face touching was observed for mask wearers (odds ratio 0.21, 95% CI 0.11-0.43, P < 0.001) and on the frequency of mucosal touches (rate ratio 0.45, 95% CI 0.29-0.69, P < 0.001). Mask wearing did not change face touching or the duration of face touches. However, significantly fewer mucosal touches were observed when wearing a mask, which may help to reduce nosocomial droplet transmission of viruses.

Data Safety Prediction Using Bird’s Eye View and Social Distancing Monitoring for Penang Roads

In terms of fatalities, Malaysia ranks third among ASEAN countries. Every year, there is an increase in accidents and fatalities. The state of the road is one factor contributing to near misses. A near miss is an almost-caused accident, an unplanned situation that could result in injury or accidents. The Majlis Bandar Pulau Pinang (MBPP) has installed 1841 closed-circuit television (CCTV) cameras around Penang to monitor traffic and track near miss incidents. When installing CCTVs, the utilisation of video allows resources to be used and optimised in situations when maintaining video memories is difficult and costly. Highways, industrial regions, and city roads are the most typical places where accidents occur. Accidents occurred at 200 per year on average in Penang from 2015 to 2017. Near misses are what create accidents. One of the essential factors in vehicle detection is the “near miss.” In this study, You Only Look Once version 3 (YOLOv3) and Faster Region-based Convolutional Neural Network (Faster RCNN) are used to solve transportation issues. In vehicle detection, a faster RCNN was used. Bird’s Eye View and Social Distancing Monitoring are used to detect the only vehicle in image processing and observe how near misses occur. This experiment tests different video quality and lengths to compare test time and error detection percentage. In conclusion, YOLOv3 outperforms Faster RCNN. In high-resolution videos, Faster RCNN outperforms YOLOv3, while in low-resolution videos, YOLOv3 outperforms Faster RCNN.

Tracking Procedural Justice in Processing Detainees: Coding Evidence from CCTV Cameras in Three Police Custody Suites

Research questionHow closely do custody suite encounters between detainees and custody suite officers (CSOs) match the procedural standards for decision makers treating people who are subject to their authority? To what degree does measurement of procedural justice displayed by CSOs on closed-circuit television (CCTV) records vary across detainees, CSOs and custody suites?DataArrest records for July, August, and September 2020 across three custody suites in the East of England were obtained and a random sample of 150 encounters selected for analysis. MethodsEncounters between CSOs and detainees at the booking-in stage as captured on pre-recorded CCTV were coded into four elements of procedural justice: voice, trustworthy motives, impartiality, and respect. Non-verbal communications and dialogue were also examined. FindingsOverall, custody suite officers demonstrated high levels of respect and neutrality in dealing with detainees. However, they showed relatively less care for the wellbeing of the detainees and did not offer them enough opportunities to ‘tell their side of the story’ (‘voice’). Further analysis revealed statistically significant variations across the three custody suites in the level of opportunities offered to detainees to have an input in discussing the decision-making. We also found evidence that as length of service as police officers and in custody roles increased, the observed level of expression of ‘trustworthy motives’ displayed decreased. Finally, detainee compliance with officers was greater when respect and care for the wellbeing of detainees were more pronounced. ConclusionsA tracking study can help identify police units and police officers with greater concentrations of procedural justice deficits. Such evidence can support more targeted training to improve the delivery of procedural justice, and enhance public confidence in policing.

Cost-Effectiveness of CCTV Surveillance Systems: Evidence from a Polish City

Closed-circuit television (CCTV) surveillance has been increasingly popular worldwide as a crime prevention measure, while its effectiveness, efficiency, and related privacy issues are debated. In this paper, we investigated whether the municipal CCTV surveillance system is cost-effective in the case study of Poznan, Poland. The analysis comprised (a) the preventive effectiveness of CCTV, (b) costs of CCTV surveillance system installation and operation, and (c) costs of crime. We collected police data on three categories of street crimes: (a) car burglary/theft, (b) car damage, and (c) robbery, as well as data on the costs of CCTV surveillance system construction and maintenance. We searched for a year after installing the CCTV system when the accumulated, hypothetical financial cost of crime was higher than the actual accumulated cost of crime and CCTV installation and maintenance in any treatment area. The number of crimes in a given category was predicted based on a comparable control area, where there were no cameras. Our study suggests that CCTV surveillance is cost-ineffective in most areas. This result implies the cautious development of CCTV surveillance, taking into account the target crime types and costs, including social costs and benefits. Evaluating cost-effectiveness, although considered as a robust measure, is in practice difficult. However, analysis of the correlation between land use characteristics and cost-effectiveness may provide practical advice on where to develop CCTV surveillance monitoring.

Surveillance Technologies and Constitutional Law.

This review focuses on government use of technology to observe, collect, or record potential criminal activity in real-time, as contrasted with "transaction surveillance" that involves government efforts to access already-existing reviews (Brayne 2018, Ridgeway 2018). Even so limited, surveillance technologies come in many guises, including closed-circuit television, automated license plate and facial readers, aerial cameras, and GPS tracking. Also classifiable as surveillance technology are devices such as thermal and electromagnetic imagers that can "see" through walls and clothing. Finally, surveillance includes wiretapping and other forms of communication interception. The following discussion briefly examines the limited evidence we have about the prevalence and effectiveness of these technologies and then describes the law governing surveillance, focusing principally on constitutional doctrine, and how it might-and might not-limit use of these technologies in the future.

Stakeholders’ attitudes towards the installations of closed-circuit television cameras in reducing school violence

PurposeWhile schools worldwide have employed closed-circuit television (CCTV) to discourage school violence, the literature shows that stakeholder attitudes toward adopting this technology are inconsistent across cultures. Generally, they are concerned with effectiveness, necessity, privacy, and operational transparency. Based on the Technology Acceptance Model for Video Surveillance (TAM-VS) theory, this study examined the attitudes of educational stakeholders, namely educators and students, toward adopting CCTV in Vietnam. This study aims to investigate which criteria Vietnamese stakeholders use to evaluate the necessity of CCTV and whether the concerns of Vietnam, a developing country, are similar to those of developed countries. MethodsThis study addressed these research questions by purposely sampling and interviewing 49 Vietnamese high school students, recent high school graduate students, teachers, and administrators. It employed semi-structured and in-depth interviews to gather content-rich data and applied thematic analysis to discover new insights from stakeholders under the modified TAM-VS framework. The study also followed the COREQ guidelines to improve the transparency and rigor of the data collection and analysis. FindingsOn an exploratory basis, most Vietnamese stakeholders believed that while CCTV cannot comprehensively prevent school violence, its effectiveness overshadows privacy concerns. Regarding operational transparency concerns, participants suggested that better communication, consent collection, and the right to amend school policies related to CCTV are important. From the participants’ perspective, the three criteria for evaluating the necessity of CCTV were (1) its effectiveness in controlling school violence, (2) serving other purposes to enhance overall schooling performance, and (3) insufficient existing methods. ConclusionsThis study provides context-rich insights into the perceptions of the necessity, effectiveness, privacy, and operational transparency concerns of CCTV across different stakeholders in the education and developing country context. On the practical contribution, the research identifies strategies that school administrators can employ to promote CCTV acceptance in Vietnamese high schools.

Human Activity Recognition: Review, Taxonomy and Open Challenges.

Nowadays, Human Activity Recognition (HAR) is being widely used in a variety of domains, and vision and sensor-based data enable cutting-edge technologies to detect, recognize, and monitor human activities. Several reviews and surveys on HAR have already been published, but due to the constantly growing literature, the status of HAR literature needed to be updated. Hence, this review aims to provide insights on the current state of the literature on HAR published since 2018. The ninety-five articles reviewed in this study are classified to highlight application areas, data sources, techniques, and open research challenges in HAR. The majority of existing research appears to have concentrated on daily living activities, followed by user activities based on individual and group-based activities. However, there is little literature on detecting real-time activities such as suspicious activity, surveillance, and healthcare. A major portion of existing studies has used Closed-Circuit Television (CCTV) videos and Mobile Sensors data. Convolutional Neural Network (CNN), Long short-term memory (LSTM), and Support Vector Machine (SVM) are the most prominent techniques in the literature reviewed that are being utilized for the task of HAR. Lastly, the limitations and open challenges that needed to be addressed are discussed.

Estimating fundamental diagram for multi-modal signalized urban links with limited probe data

Being one of the most classic concepts in the traffic flow theory, the fundamental diagram (FD) describes the relationship between average flow and average density of link-level traffic flow dynamics. Inductive loop detectors or closed-circuit television are commonly used for FD estimations and they are known to have cost-effective and accuracy issues. Thanks to the GPS-enabled smartphones and GPS-equipped probe vehicles, high temporal and spatial resolution traffic data are available which enable traffic condition inference over time and space continuously. Several existing studies have explored FD estimation algorithms on freeways where flow is generally uninterrupted and uni-modal, based on GPS trajectory data. These developments motivate this study, where the objective is to extend the application to multi-modal and interrupted environment, i.e., urban signalized areas. In this paper, an estimation method is developed to capture the FD of multi-modal traffic streams on signalized urban links. The proposed algorithm is empirically tested using real-world GPS datasets collected on a signalized arterial road in Shenzhen City. Promising results show that the proposed algorithm is capable to estimate the FD under such condition. Furthermore, impacts of multi-modal traffic and signal operations on the FD estimation are analyzed and discussed.

An Autonomous Framework for Real-Time Wrong-Way Driving Vehicle Detection from Closed-Circuit Televisions

Around 1.3 million people worldwide die each year because of road traffic crashes. There are many reasons which cause accidents, and driving in the wrong direction is one of them. In our research, we developed an autonomous framework called WrongWay-LVDC that detects wrong-way driving vehicles from closed-circuit television (CCTV) videos. The proposed WrongWay-LVDC provides several helpful features such as lane detection, correct direction validation, detecting wrong-way driving vehicles, and image capturing features. In this work, we proposed three main contributions: first, the improved algorithm for road lane boundary detection on CCTV (called improved RLB-CCTV) using the image processing technique. Second is the Distance-Based Direction Detection (DBDD) algorithm that uses the deep learning method, where the system validates and detects wrong-driving vehicles. Lastly, the Inside Boundary Image (IBI) capturing feature algorithm captures the most precise shot of the wrong-way-of-driving vehicles. As a result, the framework can run continuously and output the reports for vehicles’ driving behaviors in each area. The accuracy of our framework is 95.23%, as we tested with several CCTV videos. Moreover, the framework can be implemented on edge devices with real-time speed for functional implementation and detection in various areas.

Taxonomy of Anomaly Detection Techniques in Crowd Scenes.

With the widespread use of closed-circuit television (CCTV) surveillance systems in public areas, crowd anomaly detection has become an increasingly critical aspect of the intelligent video surveillance system. It requires workforce and continuous attention to decide on the captured event, which is hard to perform by individuals. The available literature on human action detection includes various approaches to detect abnormal crowd behavior, which is articulated as an outlier detection problem. This paper presents a detailed review of the recent development of anomaly detection methods from the perspectives of computer vision on different available datasets. A new taxonomic organization of existing works in crowd analysis and anomaly detection has been introduced. A summarization of existing reviews and datasets related to anomaly detection has been listed. It covers an overview of different crowd concepts, including mass gathering events analysis and challenges, types of anomalies, and surveillance systems. Additionally, research trends and future work prospects have been analyzed.

A Real-Time Crowdsensing Framework for Potential COVID-19 Carrier Detection Using Wearable Sensors

Artificial intelligence has been utilized extensively in the healthcare sector for the last few decades to simplify medical procedures, such as diagnosis, prognosis, drug discovery, and many more. With the spread of the COVID-19 pandemic, more methods for detecting and treating COVID-19 infections have been developed. Several projects involving considerable artificial intelligence use have been researched and put into practice. Crowdsensing is an example of an application in which artificial intelligence is employed to detect the presence of a virus in an individual based on their physiological parameters. A solution is proposed to detect the potential COVID-19 carrier in crowded premises of a closed campus area, for example, hospitals, corridors, company premises, and so on. Sensor-based wearable devices are utilized to obtain measurements of various physiological indicators (or parameters) of an individual. A machine-learning-based model is proposed for COVID-19 prediction with these parameters as input. The wearable device dataset was used to train four different machine learning algorithms. The support vector machine, which performed the best, received an F1-score of 96.64% and an accuracy score of 96.57%. Moreover, the wearable device is used to retrieve the coordinates of a potential COVID-19 carrier, and the YOLOv5 object detection method is used to do real-time visual tracking on a closed-circuit television video feed.

Using closed-circuit television cameras to analyze traffic safety at intersections based on vehicle key points detection

In the within-intersection area, vehicles from different approaches make turning movements resulting in many conflict points. Hence, drivers are more prone to make mistakes in that area, which leads to severe crash outcomes. In the current roadway system, the Closed-Circuit Television (CCTV) cameras could be a cost-effective sensor to monitor the safety condition in the within-intersection area. This study proposed a framework named “Near Miss Event Detection System (NMEDS)” for road safety diagnostics using video data collected from CCTV cameras. The proposed framework combined the Mask-RCNN bounding box detection and Occlusion-Net detection algorithm to reconstruct vehicles’ key points in a 3D view. Vehicles’ key points including right-front headlight, left-front headlight, right-back taillight, and left-back taillight could be identified and transformed into a 2D bird’s-eye view (i.e., real-world coordinate system) for safety analysis. A method was proposed to modify the occluded key points, which could not be observed by cameras due the turning movements in the within-intersection area. The post-encroachment time (PET) was calculated by using the trajectory data in the 2D view. The proposed framework was compared with two counterparts (i.e., bounding box detection only and key point detection only) by conducting an empirical study at a 4-leg intersection. The results suggested that the proposed framework could obtain more accurate vehicle trajectory and better autocorrelation analytics was conducted to identify the significantly dangerous locations in the within-intersection area. It is expected that the proposed methods could help diagnose road safety problems using CCTV cameras. Moreover, the proposed method could be incorporated with Connected Vehicle Systems and provide information to nearby drivers based on Infrastructure-to-Vehicle (I2V) technologies.