Traffic Scene Images Research Articles

Fast Traffic Sign Detection Approach Based on Lightweight Network and Multilayer Proposal Network

Vision-based traffic sign detection plays a crucial role in intelligent transportation systems. Recently, many approaches based on deep learning for traffic sign detection have been proposed and showed better performance compared with traditional approaches. However, due to difficult conditions in driving environment and the size of traffic signs in traffic scene images, the performance of deep learning-based methods on small traffic sign detection is still limited. In addition, the inference speed of current state-of-the-art approaches on traffic sign detection is still slow. This paper proposes a deep learning-based approach to improve the performance of small traffic sign detection in driving environments. First, a lightweight and efficient architecture is adopted as the base network to address the issue of the inference speed. To enhance the performance on small traffic sign detection, a deconvolution module is adopted to generate an enhanced feature map by aggregating a lower-level feature map with a higher-level feature map. Then, two improved region proposal networks are used to generate proposals from the highest-level feature map and the enhanced feature map. The proposed improved region proposal network is designed for fast and accuracy proposal generation. In the experiments, the German Traffic Sign Detection Benchmark dataset is used to evaluate the effectiveness of each enhanced module, and the Tsinghua-Tencent 100K dataset is used to compare the effectiveness of the proposed approach with other state-of-the-art approaches on traffic sign detection. Experimental results on Tsinghua-Tencent 100K dataset show that the proposed approach achieves competitive performance compared with current state-of-the-art approaches on traffic sign detection while being faster and simpler.

Predicted Anchor Region Proposal with Balanced Feature Pyramid for License Plate Detection in Traffic Scene Images

License plate detection is a key problem in intelligent transportation systems. Recently, many deep learning-based networks have been proposed and achieved incredible success in general object detection, such as faster R-CNN, SSD, and R-FCN. However, directly applying these deep general object detection networks on license plate detection without modifying may not achieve good enough performance. This paper proposes a novel deep learning-based framework for license plate detection in traffic scene images based on predicted anchor region proposal and balanced feature pyramid. In the proposed framework, ResNet-34 architecture is first adopted for generating the base convolution feature maps. A balanced feature pyramid generation module is then used to generate balanced feature pyramid, of which each feature level obtains equal information from other feature levels. Furthermore, this paper designs a multiscale region proposal network with a novel predicted location anchor scheme to generate high-quality proposals. Finally, a detection network which includes a region of interest pooling layer and fully connected layers is adopted to further classify and regress the coordinates of detected license plates. Experimental results on public datasets show that the proposed approach achieves better detection performance compared with other state-of-the-art methods on license plate detection.

Combining Unmanned Aerial Vehicles With Artificial-Intelligence Technology for Traffic-Congestion Recognition: Electronic Eyes in the Skies to Spot Clogged Roads

Unmanned aerial vehicles (UAVs) are gradually becoming useful and common. In the consumer electronics (CE) category, unmanned systems have changed the way monitoring by air is conducted in such fields as transportation, the environment, and emergency rescue. This article examines how UAVs combined with artificial-intelligence (AI) technology are used for recognizing traffic congestion. Congested roadways cause traffic delays, which wastes fuel, adds to pollution, and invites road rage. Society can benefit from technologies that help recognize and relieve traffic congestion. We present a practical framework for using UAVs to recognize traffic problems. According to this framework, images of traffic scenes are first captured by a UAV system based on route-planning technology. Then the aerial images are further processed by using convolutional neural networks (CNNs). Finally, the output is transferred to a traffic management center.

Effective Vehicle-Based Kangaroo Detection for Collision Warning Systems Using Region-Based Convolutional Networks.

Traffic collisions between kangaroos and motorists are on the rise on Australian roads. According to a recent report, it was estimated that there were more than 20,000 kangaroo vehicle collisions that occurred only during the year 2015 in Australia. In this work, we are proposing a vehicle-based framework for kangaroo detection in urban and highway traffic environment that could be used for collision warning systems. Our proposed framework is based on region-based convolutional neural networks (RCNN). Given the scarcity of labeled data of kangaroos in traffic environments, we utilized our state-of-the-art data generation pipeline to generate 17,000 synthetic depth images of traffic scenes with kangaroo instances annotated in them. We trained our proposed RCNN-based framework on a subset of the generated synthetic depth images dataset. The proposed framework achieved a higher average precision (AP) score of 92% over all the testing synthetic depth image datasets. We compared our proposed framework against other baseline approaches and we outperformed it with more than 37% in AP score over all the testing datasets. Additionally, we evaluated the generalization performance of the proposed framework on real live data and we achieved a resilient detection accuracy without any further fine-tuning of our proposed RCNN-based framework.

Cross-Domain Traffic Scene Understanding: A Dense Correspondence-Based Transfer Learning Approach

Understanding traffic scene images taken from vehicle mounted cameras is important for high-level tasks, such as advanced driver assistance systems and autonomous driving. It is a challenging problem due to large variations under different weather or illumination conditions. In this paper, we tackle the problem of traffic scene understanding from a cross-domain perspective. We attempt to understand the traffic scene from images taken from the same location but under different weather or illumination conditions (e.g., understanding the same traffic scene from images on a rainy night with the help of images taken on a sunny day). To this end, we propose a dense correspondence-based transfer learning (DCTL) approach, which consists of three main steps: 1) extracting deep representations of traffic scene images via a fine-tuned convolutional neural network; 2) constructing compact and effective representations via cross-domain metric learning and subspace alignment for cross-domain retrieval; and 3) transferring the annotations from the retrieved best matching image to the test image based on cross-domain dense correspondences and a probabilistic Markov random field. To verify the effectiveness of our DCTL approach, we conduct extensive experiments on a challenging data set, which contains 1828 images from six weather or illumination conditions.

Classify vehicles in traffic scene images with deformable part-based models

Vehicle classification is an important and challenging task in intelligent transportation systems, which has a wide range of applications. In this paper, we propose to integrate vehicle detection and vehicle classification into one single framework by using deformable part-based models. First of all, we use annotated vehicle images to train a deformable part-based model for each class of vehicles to be classified. Then, given a traffic scene image, we employ the obtained vehicle models to perform vehicle detection in it for vehicle extraction. After that, model alignment is performed on the extracted image crop, based on which features are extracted for creating a representation for the vehicle in the given image. We train a linear multi-class Support Vector Machine classifier based on representations of images in a validation set. Finally, we adopt the SVM classifier for vehicle classification. The proposed method is evaluated on the BIT-Vehicle Dataset, and can achieve an accuracy of $$91.08\%$$ , which is superior to methods used for comparison. Obtained results demonstrated the effectiveness of the proposed method.

Shape-indifferent stereo disparity based on disparity gradient estimation

A new algorithm capable of estimating disparity gradients to produce accurate dense disparities is proposed. Such a disparity gradient plays a critical role in acquiring accurate disparities for scenes with many different object shapes. The target is a road traffic scene because it contains various objects, including the road surface, vehicles, pedestrians, sidewalks, and walls. In this paper, we adopt several methods, such as initial matching cost computation, scanline optimization, left/right consistency check, and cost aggregation. However, disparity accuracy is slightly improved by the simple organization of such methods. Disparity quality decisively relies on the application of disparity gradients. Accordingly, in the proposed algorithm, cost aggregation is performed along the direction of the estimated disparity gradient in a disparity space image. This approach improves disparity quality significantly. However, this cost aggregation is time consuming. To reduce the time required, we designed a new 2D integral cost technique. The robustness of the proposed algorithm is demonstrated through the disparity maps obtained from standard images on the Web, indoor images, and outdoor images of various road traffic scenes.

Method of weak vehicle detection based on the multilevel knowledge base

A vehicle detection algorithm based on the multilevel knowledge base is proposed to overcome the problem of poor robustness as well as the difficulty of identifying weak vehicle targets. The multilevel task-driven method is adopted in the algorithm by building three different classes of knowledge bases to achieve the accuracy recognition of vehicle targets. First, a simple knowledge base is constructed via choosing Haar-like features to detect the vehicle region of interest in the traffic scene image; second, the optimal structure symmetry decision function is obtained by establishing the structure characteristics knowledge base, which is used to determine the region of the potential vehicle; finally, the property feature knowledge base is built to precisely identify vehicle targets via calculating maximum similarity. Then the relevant knowledge base will be continuously updated to achieve the method adaptive adjustment when satisfying the criterion. Experimental results illustrate that the recognition rate is more than 95% in different traffic scenarios, while the recognition rate for weak contrast vehicle targets is in excess of 71% and the false-alarm rate is simultaneously under 5%.

The View from the Driver's Seat: What Good Is Salience?

SummaryA set of experiments gauged the value of salient and non‐salient visual information to drivers' decision making. Stimuli were images of traffic scenes viewed from a drivers' approximate perspective, and the subjects' task was to make stop‐or‐go driving judgments. Under some experimental conditions (Experiment 1), subjects viewed only the salient information from within each scene. Under other conditions (Experiments 2–3), they viewed only the non‐salient information. A signal detection measure of sensitivity served as the primary dependent measure. Data indicated that an above‐chance amount of the information needed for accurate judgments was contained within salient image regions and that some information was available only from the salient regions. Other critical information, however, was contained exclusively within non‐salient regions. Results suggest that an attentional bias toward visual salience would serve as an economical information‐seeking strategy for drivers but by itself would overlook many task‐critical cues. Copyright © 2013 John Wiley & Sons, Ltd.

Attentional effects on gaze preference for salient loci in traffic scenes

Alerting drivers for self-regulation of attention might decrease crash risks attributable to absent-minded driving. However, no reliable method exists for monitoring driver attention. Therefore, we examined attentional effects on gaze preference for salient loci (GPS) in traffic scenes. In an active viewing (AV) condition requiring endogenous attention for traffic scene comprehension, participants identified appropriate speeds for driving in presented traffic scene images. In a passive viewing (PV) condition requiring no endogenous attention, participants passively viewed traffic scene images. GPS was quantified by the mean saliency value averaged across fixation locations. Results show that GPS was less during AV than during PV. Additionally, gaze dwell time on signboards was shorter for AV than for PV. These results suggest that, in the absence of endogenous attention for traffic scene comprehension, gaze tends to concentrate on irrelevant salient loci in a traffic environment. Therefore, increased GPS can indicate absent-minded driving. Practitioner Summary: The present study demonstrated that, without endogenous attention for traffic scene comprehension, gaze tends to concentrate on irrelevant salient loci in a traffic environment. This result suggests that increased gaze preference for salient loci indicates absent-minded driving, which is otherwise difficult to detect.

The Design of the Detection and Localization of the Color Image for the Location of Road Traffic Signs

An intelligent recognition algorithm of the combination of image edge detection and image area segmentation is designed in this paper. The image edge detection adopts the edge detection method of color feature space. The color multi-threshold region segmentation method is used the image area segmentation. The algorithm is mainly used for the detection and location of city road traffic signs. Some traffic scene images are processed by this algorithm. The simulation results show that the combination algorithm has better segmentation ability. It is helpful to realize the intelligent recognition of road traffic signs.

The Effect of Simulated Cataracts on Drivers’ Hazard Perception Ability

We investigated the extent to which simulated cataracts slow a driver's ability to anticipate potential traffic hazards, a skill that has been found to correlate with crash risk. In previous studies, we found a significant correlation between contrast sensitivity and hazard perception in a sample of older drivers. The present study allowed us to determine the causal direction of this relationship. This is important as it provides a better understanding of the mechanisms underlying the higher crash risk of drivers with cataracts. One hundred eighty-six drivers with normal vision completed a validated video-based hazard perception driving test, designed to measure hazard anticipation response times in dynamic scenes. They also completed a change detection task based on traffic hazards, which was designed to measure object detection times in static scenes. Participants undertook the tasks wearing either mild or moderate cataract simulation goggles, or wearing goggle frames without lenses. Participants wearing moderate simulated cataract goggles were slower than the control group in both the hazard perception test, t(98.50) = -3.71, p < 0.001, and the hazard change detection task, t(124) = -13.86, p < 0.001. Participants with the mild simulated cataract goggles were slower than the control group in the hazard change detection task, t(114) = -4.04, p < 0.001, but not the hazard perception test, t(114) = -1.33, p = 0.19. Moderate levels of simulated cataract slowed drivers' ability to detect and anticipate traffic hazards enough to warrant road safety concerns, despite the fact that the vision of participants wearing the cataract goggles still complied with the minimum legal standard required for driving.

Age Differences in Visual Search for Traffic Signs During a Simulated Conversation

The effects of divided attention were examined in younger adults (M = 23 years) and older adults (M = 64 years) who searched for traffic signs in digitized images of traffic scenes. Sign search was executed under single-task and dual-task conditions in scenes containing either small or large amounts of visual clutter. For both age groups, clutter and the secondary task had additive effects on search accuracy, speed, and oculomotor involvement. Compared with the younger adults, older adults were less accurate, especially with high-clutter scenes, were slower to decide that a target sign was not present, and exhibited a marginally greater divided-attention effect on reaction times. They exhibited longer fixations in the divided-attention condition, in which they also showed a disproportionate reduction in recognition memory for the content of the secondary task. Actual or potential applications of this research include methods for evaluating the distraction of conversations and safety implications of conversation on visual search behavior.

Characteristics of Passing and Meeting Maneuvers on Exclusive Bicycle Paths

The findings are presented of a comprehensive analysis of the characteristics of passing and meeting maneuvers of bicycles traveling on an exclusive bicycle path 3 m (10 ft) wide. The study is based on bicycle location data collected every 0.5 s from video images of bicycle traffic scenes recorded in Denver, Colorado. For passing maneuvers, the average speeds of passing and passed bicycles, the relationship between passing and passed bicycle speeds every 0.5 s, comfortable speed difference threshold, lateral spacing during passing, average passing lengths, and the shape of passing are reported. For meeting maneuvers, statistics about the average speed and lateral spacing are included; for paired riding, headway and lateral spacing statistics are included. The results also show that the average bicycle speeds estimated are significantly higher than the speeds reported from a study conducted in the Netherlands. These findings can be used to develop more accurate procedures for the analysis of the quality of service of bicycle paths. According to the 2000 Highway Capacity Manual (HCM), a weighted sum of the number of passing and meeting events is used as the criterion to determine the level of service for bicycle paths. The procedures in the 2000 HCM are based on research mainly conducted in the Netherlands for two-lane or 2.4-m-wide bicycle paths and have been extended to 3-m-wide paths. The findings presented here are for 3-m-wide bicycle paths in the United States and therefore may be used to develop new procedures for HCM as well as provide the data needed to develop a bicycle simulation model.

Model-Based Localisation and Recognition of Road Vehicles

Objects are often constrained to lie on a known plane. This paper concerns the pose determination and recognition of vehicles in traffic scenes, which under normal conditions stand on the ground-plane. The ground-plane constraint reduces the problem of localisation and recognition from 6 dof to 3 dof. The ground-plane constraint significantly reduces the pose redundancy of 2D image and 3D model line matches. A form of the generalised Hough transform is used in conjuction with explicit probability-based voting models to find consistent matches and to identify the approximate poses. The algorithms are applied to images of several outdoor traffic scenes and successful results are obtained. The work reported in this paper illustrates the efficiency and robustness of context-based vision in a practical application of computer vision. Multiple cameras may be used to overcome the limitations of a single camera. Data fusion in the proposed algorithms is shown to be simple and straightforward.

Locating text in complex color images

There is a substantial interest in retrieving images from a large database using the textual information contained in the images. An algorithm which will automatically locate the textual regions in the input image will facilitate this task; the optical character recognizer can then be applied to only those regions of the image which contain text. We present two methods for automatically locating text in complex color images. The first method segments the image into connected components with uniform color, and uses several heuristics (size, alignment, proximity) to select the components which are likely to contain character(s) belonging to the text. The second method computes the local spatial variation in the gray-scale image, and locates text in regions with high variance. A combination of the two approaches is shown to be more effective than the individual methods. The proposed methods have been used to locate text in compact disc (CD) and book cover images, as well as in the images of traffic scenes captured by a video camera. Initial results are encouraging and suggest that these algorithms can be used in image retrieval applications.

Road Traffic Monitoring Using Image Processing—A Survey of Systems, Techniques and Applications

Following their earlier research work associated with the development of the TRIP System at Sheffield University/UMIST the authors are engaged in a re-evaluation of image processing applications in transportation engineering. Existing image processing techniques applied to traffic data collection involve only low level image interpretation of sequences of images. Although primitive these techniques have enabled useful parameters such as vehicle speed, counts and lane occupancy to be measured effectively. It is believed, however, that future image processing systems will require additional built-in intelligence to cope with more sophisticated applications.This paper will provide a review of the experience of a number of groups active in image processing research in the transport field. It will go on to comment upon the short-falls of existing equipment and processing techniques. These considerations were taken into account during the development of the TRIP II system, which will then be described. The paper will go on to suggest methods whereby some of the inadequacies of past systems may be overcome in the light of trends in computing equipment, special purpose hardware and enhanced processing technique and machine intelligence.In order to achieve near real-time operation past systems have concentrated upon applying differencing techniques to small samples of video images. Conventional systems have often been based on purpose built computers and application specific algorithms. However the TRIP II system takes full advantage of the area sensing capability of video images. The approach is different since the hardware is based upon an inexpensive IBM PC/AT microcomputer. The software provides an alternative means of processing complex pattern information by simulating the operation of a neural network. The vehicle detection algorithm relies on the ability of the TRIP II system to learn from examples and discriminate between complex patterns within video images of traffic scenes. Over the past decade many image processing systems for traffic monitoring have been designed and implemented. The paper will finally go on to outline the range of applications for such image analysis systems.