Dashcam Videos Research Articles

Stationary and Moving Camera Video Analysis Compared to Known Reference System

<div class="section abstract"><div class="htmlview paragraph">Video recordings of vehicular collisions have become widely available to the accident reconstructionist and can play a vital role in determining the locations and speeds of the subject vehicles involved in a collision. However, due to varying video resolutions, framerate, lens distortion, motion blur, and camera movement, errors in video analysis can occur. To understand the total error inherent to video analysis, this study presents analysis of videos from different video systems, the limitations in the analysis, and a comparison of video analysis speeds to reference datasets. The factors that influenced variance included resolution, lens correction, shutter speed, and framerate. The video systems analyzed included three moving cameras and two stationary units.</div><div class="htmlview paragraph">In the present study, a mock collision scenario in which a target vehicle approached a recording vehicle head-on, was staged to emulate an actual event captured on video. The target vehicle’s speed was analyzed using the captured videos. The video analysis results were then compared to the speeds obtained from the target vehicle’s wheel speed sensors via a VBOX system connected to the vehicle’s Controlled Area Network (CAN), and VBOX GPS position data.</div><div class="htmlview paragraph">The videos were recorded from two locations. Location 1 was at the top corner of a business complex. This location was equipped with two video cameras: a GoPro HERO5 and a Sony α6400 mirrorless camera. Location 2 was within the recording vehicle itself. This location was equipped with three video systems: a 2018 Tesla Model 3 Dashcam video camera system, a generic dashboard video system with a low framerate, and a Blackmagic Design camera. Videos were captured as the recording vehicle moved towards the target vehicle; the camera’s location and angle relative to the target were constantly changing.</div><div class="htmlview paragraph">Errors in the determined speeds were quantified based on comparison of the video analysis speeds to the reference datasets. The errors in speed were determined to be inversely correlated to the video resolution. Additionally, the analysis of the video footage from the stationary source yielded lower error than the analysis of the moving vehicle video for a given resolution and framerate.</div></div>

Vehicle Speed Measurement Methodology Robust to Playback Speed-Manipulated Video File

Concomitant with the virtual ubiquity of dashcams in vehicles, the instantaneous speed of vehicles during accidents can now be measured using dashcam videos to understand the causes of such accidents. In this study, we estimated vehicle speeds using dashcam videos recorded in various road environments and analyzed them by comparing them to ground truth. We also considered various scenarios yielding retaken and re-encoded videos from additional recording devices, depending on the circumstances, wherein the original video can be manipulated in terms of playback speed and/or frames per second. The experimental results revealed that the actual and measured speeds were similar for most user-specific intervals. Moreover, the results suggest that our estimated speed is adequately valid as evidence, regardless of retake and re-encoding states. Furthermore, we adopted a statistical approach (i.e., Pearson’s correlation coefficient analysis) to verify the proposed vehicle speed estimation method. The results showed that when playback speed was manipulated in re-encoded or retaken video, the proposed method consistently measured speeds at 99.99 % accuracy. This verified method will play an important role in deriving reliable results from various evidential video sources.

Vehicle-Level Traffic Accident Detection on Vehicle-Mounted Camera Based on Cascade Bi-LSTM

In this paper, we propose a traffic accident detection on vehicle-mounted camera. In the proposed method, the minimum bounding box coordinates the central coordinates on the bird"s eye view and motion vectors of each vehicle object, and ego-motions of the vehicle equipped with dash-cam are extracted from the dash-cam video. By using extracted 4 kinds features as the input of Bi-LSTM (bidirectional LSTM), the accident probability (score) is predicted. To investigate the effect of each input feature on the probability of an accident, we analyze the performance of the detection the case of using a single feature input and the case of using a combination of features as input, respectively. And in these two cases, different detection models are defined and used. Bi-LSTM is used as a cascade, especially when a combination of the features is used as input. The proposed method shows 76.1% precision and 75.6% recall, which is superior to our previous work.

Automatic, location-privacy preserving dashcam video sharing using blockchain and deep learning

Today, many people use dashcams, and videos recorded on dashcams are often used as evidence of accident fault. People can upload videos of dashcam recordings with specific accident clips and share the videos with others who request them, by providing the time or location of an accident. However, dashcam videos are erased when the dashcam memory is full, so periodic backup is necessary for video sharing. It is inconvenient for dashcam owners to search for and transmit a requested video clip from backup videos. In addition, anonymity is not ensured, which may reduce location privacy by exposing the video owner’s location. To solve this problem, we propose a video sharing scheme with accident detection using deep learning coupled with automatic transfer to the cloud; we also propose ensuring data and operational integrity along with location privacy by using blockchain smart contracts. Furthermore, our proposed system uses proxy re-encryption to enhance the confidentiality of a shared video. Our experiments show that our proposed automatic video sharing system is cost-effective enough to be acceptable for deployment.

Risk perception: A study using dashcam videos and participants from different world regions

Objective: Research has shown that perceived risk is a vital variable in the understanding of road traffic safety. Having experience in a particular traffic environment can be expected to affect perceived risk. More specifically, drivers may readily recognize traffic hazards when driving in their own world region, resulting in high perceived risk (the expertise hypothesis). Oppositely, drivers may be desensitized to traffic hazards that are common in their own world region, resulting in low perceived risk (the desensitization hypothesis). This study investigated whether participants experienced higher or lower perceived risk for traffic situations from their region compared to traffic situations from other regions. Methods: In a crowdsourcing experiment, participants viewed dashcam videos from four regions: India, Venezuela, United States, and Western Europe. Participants had to press a key when they felt the situation was risky. Results: Data were obtained from 800 participants, with 52 participants from India, 75 from Venezuela, 79 from the United States, 32 from Western Europe, and 562 from other countries. The results provide support for the desensitization hypothesis. For example, participants from India perceived low risk for hazards (e.g., a stationary car on the highway) that were perceived as risky by participants from other regions. At the same time, support for the expertise hypothesis was obtained, as participants in some cases detected hazards that were specific to their own region (e.g., participants from Venezuela detected inconspicuous roadworks in a Venezuelan city better than did participants from other regions). Conclusion: We found support for the desensitization hypothesis and the expertise hypothesis. These findings have implications for cross-cultural hazard perception research.

Prediction of effort and eye movement measures from driving scene components

For transitions of control in automated vehicles, driver monitoring systems (DMS) may need to discern task difficulty and driver preparedness. Such DMS require models that relate driving scene components, driver effort, and eye measurements. Across two sessions, 15 participants enacted receiving control within 60 randomly ordered dashcam videos (3-second duration) with variations in visible scene components: road curve angle, road surface area, road users, symbols, infrastructure, and vegetation/trees while their eyes were measured for pupil diameter, fixation duration, and saccade amplitude. The subjective measure of effort and the objective measure of saccade amplitude evidenced the highest correlations (r = 0.34 and r = 0.42, respectively) with the scene component of road curve angle. In person-specific regression analyses combining all visual scene components as predictors, average predictive correlations ranged between 0.49 and 0.58 for subjective effort and between 0.36 and 0.49 for saccade amplitude, depending on cross-validation techniques of generalization and repetition. In conclusion, the present regression equations establish quantifiable relations between visible driving scene components with both subjective effort and objective eye movement measures. In future DMS, such knowledge can help inform road-facing and driver-facing cameras to jointly establish the readiness of would-be drivers ahead of receiving control.

Progressive Temporal-Spatial-Semantic Analysis of Driving Anomaly Detection and Recounting.

For analyzing the traffic anomaly within dashcam videos from the perspective of ego-vehicles, the agent should spatial-temporally localize the abnormal occasion and regions and give a semantically recounting of what happened. Most existing formulations concentrate on the former spatial-temporal aspect and mainly approach this goal by training normal pattern classifiers/regressors/dictionaries with large-scale availably labeled data. However, anomalies are context-related, and it is difficult to distinguish the margin of abnormal and normal clearly. This paper proposes a progressive unsupervised driving anomaly detection and recounting (D&R) framework. The highlights are three-fold: (1) We formulate driving anomaly D&R as a temporal-spatial-semantic (TSS) model, which achieves a coarse-to-fine focusing and generates convincing driving anomaly D&R. (2) This work contributes an unsupervised D&R without any training data while performing an effective performance. (3) We novelly introduce the traffic saliency, isolation forest, visual semantic causal relations of driving scene to effectively construct the TSS model. Extensive experiments on a driving anomaly dataset with 106 video clips (temporal-spatial-semantically labeled carefully by ourselves) demonstrate superior performance over existing techniques.

Dehumanizing Slave Personhood

AbstractAfrohumanism is crucial to the forward-looking “project of thinking humanity from perspectives beyond the liberal humanist subject, Man” (Weheliye 2014: 8). It is another question, however, whether such a humanist approach provides the best historical analytic for understanding slavery and its carceral afterlives. This question becomes particularly pressing when we consider that today’s prison-industrial complex, like the American slaveholder of the past, extracts profits by strategically exploiting—rather than denying—the lucrative humanity of its captive black and brown subjects. To illustrate these claims, this article examines a seldom-discussed slave case, United States v. Amy (1859), which was tried before Supreme Court chief justice Roger B. Taney two years after his infamous decision in Dred Scott v. Sandford (1857). Centering on the figure of the legal person rather than the human or the citizen, United States v. Amy alerts us to the lethal legacy of slave personhood as a debilitating mixture of civil death and criminal culpability. Nowhere, perhaps, is that legacy more evident than in viral videos of police misconduct. And nowhere do we see a more vivid assertion of black counter-civility than in the dash cam video of the late Sandra Bland’s principled, outraged response to her pretextual traffic stop by Trooper Brian Encinia. The essay closes by considering Bland’s arrest and subsequent death in custody in the context of her own and other African Americans’ efforts to achieve and maintain a civil presence in an American law and culture where black personhood remains legible primarily as criminality.

Egocentric Analysis of Dash-Cam Videos for Vehicle Forensics

Video acquisition using dashboard-mounted cameras has recently achieved massive popularity around the world. One of the major developments following the dash-cam's popularity is that videos captured by them can be used as testimony during scenarios, like traffic violations and accidents. The widespread deployment of dash-cams brings new problems ranging from the compromise of privacy by uploading these videos on public websites using videos captured from other cars for making fraudulent claims. Therefore, there is a compelling need to address the problems associated with the usage of dash-cam videos. In this paper, we discuss and highlight the importance of the emerging area of multimedia vehicle forensics. We propose an algorithm for linking a dash-cam video to a specific car. The proposed algorithm is useful for various applications, for example, insurance companies can authenticate the origin of video before processing the claim. In a different scenario of illegitimate video upload on the Web, the video can be traced back to the car it originated from. To this end, we make use of motion blur extracted from dash-cam videos for generating a discriminative feature. We observe that the subtle motion pattern of every vehicle can serve as its unique signature. We extract motion blur from dash-cam videos and use random forest trees for classifying the vehicle correctly. The experimental results on thousands of frames obtained from dash-cam videos of several cars show the effectiveness of our approach. We further investigate the process of forging the signature of a car and propose a counter forensics method to detect such forgery. Also, we discuss the application of our technique to other potential platforms where the camera can be mounted, for example, on the chest of a person. We believe that ours is the first work that describes this new area of research.

Early Detection of Vacant Parking Spaces Using Dashcam Videos

A major problem in metropolitan areas is finding parking spaces. Existing parking guidance systems often adopt fixed sensors or cameras that cannot provide information from the driver’s point of view. Motivated by the advent of dashboard cameras (dashcams), we develop neural-network-based methods for detecting vacant parking spaces in videos recorded by a dashcam. Detecting vacant parking spaces in dashcam videos enables early detection of spaces. Different from conventional object detection methods, we leverage the monotonicity of the detection confidence with respect to the distance away of the approaching target parking space and propose a new loss function, which can not only yield improved detection results but also enable early detection. To evaluate our detection method, we create a new large dataset containing 5,800 dashcam videos captured from 22 indoor and outdoor parking lots. To the best of our knowledge, this is the first and largest driver’s view video dataset that supports parking space detection and provides parking space occupancy annotations.

Body camera footage leads to lower judgments of intent than dash camera footage

Police departments use body-worn cameras (body cams) and dashboard cameras (dash cams) to monitor the activity of police officers in the field. Video from these cameras informs review of police conduct in disputed circumstances, often with the goal of determining an officer's intent. Eight experiments (N = 2,119) reveal that body cam video of an incident results in lower observer judgments of intentionality than dash cam video of the same incident, an effect documented with both scripted videos and real police videos. This effect was due, in part, to variation in the visual salience of the focal actor: the body cam wearer is typically less visually salient when depicted in body versus dash cam video, which corresponds with lower observer intentionality judgments. In showing how visual salience of the focal actor may introduce unique effects on observer judgment, this research establishes an empirical platform that may inform public policy regarding surveillance of police conduct.

Dash Cam videos on YouTube™ offer insights into factors related to moose-vehicle collisions

To gain a better understanding of the dynamics of moose-vehicle collisions, we analyzed 96 videos of moose-vehicle interactions recorded by vehicle dash-mounted cameras (Dash Cams) that had been posted to the video-sharing website YouTube™. Our objective was to determine the effects of road conditions, season and weather, moose behavior, and driver response to actual collisions compared to near misses when the collision was avoided. We identified 11 variables that were consistently observable in each video and that we hypothesized would help to explain a collision or near miss. The most parsimonious logistic regression model contained variables for number of moose, sight time, vehicle slows, and vehicle swerves (AICcw = 0.529). This model had good predictive accuracy (AUC = 0.860, SE = 0.041). The only statistically significant variable from this model that explained the difference between moose-vehicle collisions and near misses was ‘Vehicle slows’. Our results provide no evidence that road surface conditions (dry, wet, ice or snow), roadside habitat type (forested or cleared), the extent to which roadside vegetation was cleared, natural light conditions (overcast, clear, twilight, dark), season (winter, spring and summer, fall), the presence of oncoming traffic, or the direction from which the moose entered the roadway had any influence on whether a motorist collided with a moose. Dash Cam videos posted to YouTube™ provide a unique source of data for road safety planners trying to understand what happens in the moments just before a moose-vehicle collision and how those factors may differ from moose-vehicle encounters that do not result in a collision.

Pothole detection using spatio‐temporal saliency

This study proposes a simple and fast pothole detection method of a video using spatio-temporal saliency, where a pothole is a type of failure on an asphalt pavement. In monocular vision, potholes typically have two visual properties. First, potholes have low-intensity areas that are darker than nearby pavement because of shadows. Second, the texture inside the potholes is coarser than the nearby pavement. However, these two visual properties are not sufficient for accurately detecting potholes because of shadows on the pavement and lack of the texture. In this study, the above challenges are addressed by detecting slower objects that are coming closer to a moving vehicle. Pothole is a stationary object ahead on the road. In terms of the relative speed, pothole is a slower object than a moving vehicle. In the video sequence acquired from a dash-cam, different image characteristics appear depending on the relative speed of objects such as potholes and cars ahead. The proposed method detects outliers using spatio-temporal saliency in the yt image: a slower object is represented as diagonal outliers. Experiments with several dash-cam videos show the effectiveness of the proposed method for real video sequences that contain potholes, cars, road lanes, and pavement markings.

Introduction: A Special Issue of <em>Spectrum: A Journal on Black Men</em>

A Special Issue of Spectrum:A Journal on Black Men Terrell L. Strayhorn With so much going on in the country, the need for this special issue of Spectrum: A Journal on Black Men is crystal clear and compelling. From the tragic killing of unarmed 17-year-old Trayvon Martin (5 feet 11 inches, 158 pounds) on February 26, 2012 in Sanford, Florida, at the hands of George Zimmerman to the fatal choke-hold that killed unarmed 43-year-old Eric Garner (6 feet 3 inches, 350 pounds) on July 17, 2014 in Staten Island, New York, at the hands (and arms) of a gang of police officers, the message is loud and obvious that race [still] matters (West, 1993), manhood matters, and Black men—no matter how tall or short, thick or thin, educated or illiterate, rich or poor—face an uncertain future in this country. Without intervention, Black men in the United States are far more likely to be killed at the hands (or arms) of lawless White male avengers masquerading as “good cops” than to graduate high school, complete college, vote in an election, run for public office, or earn gainful employment. Contrary to this swelling tide of national tragedies and startling statistics that suggest otherwise, Black men matter in all sectors of society, especially education. The headlines and stories that provide clear evidence about the need for this special issue of Spectrum: A Journal on Black Men go beyond the tragic loss of Trayvon Martin in Florida and Eric Garner in New York. Many will recall the fatal shooting of 18-year-old Michael Brown (6 feet 4 inches, 292 pounds) on August 9, 2014 in Ferguson, Missouri, that sparked enraged protests and a groundswell of political actions initiating the contemporary “#BlackLivesMatter” movement. Four days prior (August 5, 2014), 22-year-old John Crawford, III, was killed by Beavercreek Officer Sean Williams while holding a toy BB gun in a Walmart near Dayton, Ohio. Then, 12-year-old Tamir Elijah Rice (5 foot 7 inches, 195 pounds) was killed by Officers Frank Garmback and Timothy Loehmann on November 22, 2014 in Cleveland, Ohio’s Cudell Recreation Center Park—both of these tragic incidents occured in the state of Ohio where I currently live and work. [End Page 1] Sadly, countless other incidents abound—and, of course, these are only the “reported” cases that make national headline news (so many more are never reported or given prime time). For instance, news broke again recently (November 2015) after a police dashcam video was released to the public showing the brutal murder of 17-year-old Laquan McDonald on October 20, 2014 in Chicago, Illinois, at the bloodied hands of Officer Jason Van Dyke. And thanks to Feidin Santana, who turned over his smartphone-recorded video footage, we now know the real truth (emphasis added) about the last living moments of 50-year-old Walter Scott, who was gunned down from the back by Officer Michael Slager on April 4, 2015 in North Charleston, South Carolina. Eight days later, 25-year-old Freddie Carlos Gray was arrested by six Baltimore officers, thrown around in the back of a police van, and died from spinal injuries on April 19, 2015. Tragic. Devastating. Exhausting. It would be easy to throw up one’s hands in surrender to the threatening message that these events issue—namely, that Black men’slives don’t matter and they can be killed for any reason at any time (and then blamed for their own victimization). But we at Spectrum: A Journal on Black Men firmly believe that Black men matter in all sectors of society and our country lives beneath its collective potential without them. The tragic stories don’t stop just because I have “put a period” (.) on my temporally ordered summary of these horrific episodes. In fact, each draft of this special issue introduction quickly becomes outdated as news breaks daily about the deplorable situation of Black men in the United States. What I offer here is nowhere close to an encyclopedic review of all #BlackLivesMatter cases or all shootings of Black men by uniformed law officials—that task goes way beyond the purpose...