Challenging Weather Conditions Research Articles

Adaptive Multi-Pedestrian Tracking by Multi-Sensor: Track-to-Track Fusion Using Monocular 3D Detection and MMW Radar

Accurate and reliable tracking of multi-pedestrian is of great importance for autonomous driving, human-robot interaction and video surveillance. Since different scenarios have different best-performing sensors, sensor fusion perception plans are believed to have complementary modalities and be capable of handling situations which are challenging for single sensor. In this paper, we propose a novel track-to-track fusion strategy for multi-pedestrian tracking by using a millimeter-wave (MMW) radar and a monocular camera. Pedestrians are firstly tracked by each sensor according to the sensor characteristic. Specifically, the 3D monocular pedestrian detections are obtained by a convolutional neural network (CNN). The trajectory is formed by the tracking-by-detection approach, combined with Bayesian estimation. The measurement noise of the 3D monocular detection is modeled by a detection uncertainty value obtained from the same CNN, as an approach to estimate the pedestrian state more accurately. The MMW radar utilizes the track-before-detection method due to the sparseness of the radar features. Afterwards, the pedestrian trajectories are obtained by the proposed track-to-track fusion strategy, which can work adaptively under challenging weather conditions, low-illumination conditions and clutter scenarios. A group of tests are carried out to validate our pedestrian tracking strategy. Tracking trajectories and optimal sub-pattern assignment (OSPA) metric demonstrate the accuracy and robustness of the proposed multi-sensor multi-pedestrian tracking system.

Moving Object Detection in Video with Algorithms YOLO and Faster R-CNN in Different Conditions

In this study, YOLOv4 with YOLOv3 and Faster R-CNN compared for object detection both under challenging weather conditions and in the dark. It is difficult to detect moving objects such as pedestrians, cars, buses, and motorcycles in bad weather conditions or at night, especially in adverse weather conditions such as rain, fog, and snow. The objective of this study is to assess the performance of these three algorithms in such circumstances, as none of them were designed to work in bad weather or at night. Tesla P4 GPUs with 12GB of RAM were used for this study, with algorithms trained using open-image datasets, where YOLOv4 had the highest performance at 40,000 iterations, 72% mAP, and 63% recall. While YOLOv3 has achieved maximum at 36000 iterations, 65.53% mAP, and 54% recall, Faster R-CNN has achieved maximum at 36,000 iterations, 51% mAP, and 49% recall. According to the results, YOLOv4 performed the best compared to YOLOv3 and Faster R-CNN.

DIOR: A Hardware-Assisted Weather Denoising Solution for LiDAR Point Clouds

The interest in developing and deploying fully autonomous vehicles on our public roads has come to a full swing. Driverless capabilities, widely spread in modern vehicles through advanced driver-assistance systems (ADAS), require highly reliable perception features to navigate the environment, being light detection and ranging (LiDAR) sensors a key instrument in detecting the distance and speed of nearby obstacles and in providing high-resolution 3D representations of the surroundings in real-time. However, and despite being assumed as a game-changer in the autonomous driving paradigm, LiDAR sensors can be very sensitive to adverse weather conditions, which can severely affect the vehicle’s perception system behavior. Aiming at improving the LiDAR operation in challenging weather conditions, which contributes to achieving higher driving automation levels defined by the Society of Automotive Engineers (SAE), this article proposes a weather denoising method called Dynamic light-Intensity Outlier Removal (DIOR). DIOR combines two approaches of the state-of-the-art, the dynamic radius outlier removal (DROR) and the low-intensity outlier removal (LIOR) algorithms, supported by an embedded reconfigurable hardware platform. By resorting to field-programmable gate array (FPGA) technology, DIOR can outperform state-of-the-art outlier removal solutions, achieving better accuracy and performance while guaranteeing the real-time requirements.

Soft Warping Based Unsupervised Domain Adaptation for Stereo Matching

Stereo matching is a practical method to estimate depth information and retrieve 3D world in robot perception and autonomous driving scenarios. With the development of convolution neural networks (CNNs), deep-learning based stereo matching algorithms have significantly improved the accuracy and dominated most of the online benchmarks. However, limited labels in real world, especially in challenging weather conditions, still hinder the technology from practical usage. In this paper, we propose a new unsupervised learning mechanism for stereo matching, utilizing adversarial iterative learning and novel soft warping loss to promote the effectiveness of the networks in unseen environments. The experiments transferring the stereo matching module from synthetic domain to real-world domain demonstrate the superiority of our proposed method. Extensive experiments in challenging weathers further prove that our method shows great practical potential in strait environments.

Review on Lane Detection and Tracking Algorithms of Advanced Driver Assistance System

Autonomous vehicles and advanced driver assistance systems are predicted to provide higher safety and reduce fuel and energy consumption and road traffic emissions. Lane detection and tracking are the advanced key features of the advanced driver assistance system. Lane detection is the process of detecting white lines on the roads. Lane tracking is the process of assisting the vehicle to remain in the desired path, and it controls the motion model by using previously detected lane markers. There are limited studies in the literature that provide state-of-art findings in this area. This study reviews previous studies on lane detection and tracking algorithms by performing a comparative qualitative analysis of algorithms to identify gaps in knowledge. It also summarizes some of the key data sets used for testing algorithms and metrics used to evaluate the algorithms. It is found that complex road geometries such as clothoid roads are less investigated, with many studies focused on straight roads. The complexity of lane detection and tracking is compounded by the challenging weather conditions, vision (camera) quality, unclear line-markings and unpaved roads. Further, occlusion due to overtaking vehicles, high-speed and high illumination effects also pose a challenge. The majority of the studies have used custom based data sets for model testing. As this field continues to grow, especially with the development of fully autonomous vehicles in the near future, it is expected that in future, more reliable and robust lane detection and tracking algorithms will be developed and tested with real-time data sets.

PSXI-17 Performance of beef cattle overwintered on bale-grazed pasture or in a dry lot in south-central North Dakota

Abstract The performance of beef cows managed in two overwintering environments, bale-grazed pasture or dry lot pens, was evaluated in a study conducted over four winters, from 2016 to 2019. Non-lactating pregnant Angus cows (2016, n = 32, body weight [BW] = 599 ± 68 kg; 2017, n = 40, BW = 620 ± 59 kg; 2018, n = 40, BW = 643 ± 47; 2019, n = 40, BW = 624 ± 30) were divided into four groups of similar body weight and randomly assigned to either bale grazing paddocks or dry lot pens. Cows in both housing scenarios were fed the same hay (7.5% CP; 51.7% TDN). Two-day body weights were taken at the start and end of the study. Two independent observers assigned body condition score (BCS) using a 9-point system (1 = emaciated, 9 = obese at the start and end of the study. Keeping cows on pasture or in dry lot pens did not influence (P > 0.05) final BW and BCS. However, daily gains and BCS change were greater (P ≤ 0.05) in bale-grazed cows relative to cows kept in dry lot pens. Birth weights, weaning weights and daily gains of calves from cows kept in the two overwintering environments were similar (P > 0.05). This study suggests that there are no advantages to keeping cows in dry lot pens in winter as long as cows are protected from wind and have adequate access to water and feed. In certain situations, such as in blizzards, it may be an advantage to keep cows on pasture where they can access feed since challenging weather conditions can prevent access to cows in dry lot pens.

Pose estimation at night in infrared images using a lightweight multi-stage attention network

Human Keypoints Detection is a relatively basic task in computer vision; it is the pre-task of human action recognition, behavior analysis and human–computer interaction. Since most abnormal actions occur at night, how to effectively extract skeleton sequence data in a low-light or completely dark environment poses a huge challenge for its identification. This paper proposes to use far infrared images to detection key points of the human body, which can solve the problem of human pose estimation under challenging weather conditions such as total darkness, smoke, inclement weather and glare. However, far-infrared images have some shortcomings, such as low resolution, noise and thermal characteristics; the skeleton data need to be provided in real time for the next stage of task. Based on the above reasons, this paper proposes a lightweight multi-stage attention network (LMANet) to detect the key points of human at night. This new network structure adds context information through the large receptive field, which helps to assist the detection of neighboring key points through this information, but for the sake of lightweight consideration, this article only extends the network to two stages. In addition, this article uses the attention module to effectively select channels with a large amount of information and highlight the features of key points, while eliminating background interference. In order to detect key points of the human in various complex environments, we use techniques such as difficult sample mining which improves the accuracy of key points with low confidence. Our network has been verified on two visible light datasets, fully demonstrating excellent performance. This paper successfully introduces far-infrared images into the field of pose estimation, because there is no public dataset for far-infrared pose estimation. In this paper, 700 images are selected for annotation from multiple public far-infrared object detection, segmentation and action recognition datasets; our algorithm is verified on this dataset; the effect is very good. After the paper is published, we will publish our key points of the human body annotated documents.

Information and Communication Technologies Usage in Working with Families of Early Aged and Preschool Children with Visual Impairments in Terms of Distance Psychological and Pedagogical Support

This paper presents the results of studying the use of information and communication technologies (ICT) in working with families of early aged and preschool children with visual impairments. The authors analyze the current situation in families during strict quarantine in 2020 and 2021 and study the possibilities of using information technology in the development of children with visual impairments at home. The study describes how or under what circumstances specialists worked with families of early aged and preschool children with visual impairments, what problems parents, professionals faced, and whether they were ready to solve—via the author’s questionnaires for professionals and parents, identified current problems that need urgent solutions. The theoretical basis and practical side of studying the problem of using information and communication technologies in families’ distance psychological and pedagogical support are presented. Attention is focused on different levels, and specialists with children with visual impairments in the context of family education and the absence or insufficiency of distance correction work with early aged children and their parents are emphasized. The need for further study of the situation for the development of appropriate forms, methods of ICT in working with families of children of early and preschool age with visual impairments. The relevance of scientific research on the use of ICT and distance psychological and pedagogical support is proved, as this format will be in demand not only during lockdown but also in seasonal outbreaks, challenging weather conditions, as well as constantly for families living in settlements, remote from International Rescue Committee and other educational institutions, families who due to various life circumstances are not able to visit institutions that provide educational and psychological services. It is concluded that there is an urgent need to create a holistic system using ICT to work with families of early aged and preschool children with visual impairments, which will help solve a wide range of psychological and pedagogical rehabilitation support tasks.

Lightweight Object Detection Ensemble Framework for Autonomous Vehicles in Challenging Weather Conditions.

The computer vision systems driving autonomous vehicles are judged by their ability to detect objects and obstacles in the vicinity of the vehicle in diverse environments. Enhancing this ability of a self-driving car to distinguish between the elements of its environment under adverse conditions is an important challenge in computer vision. For example, poor weather conditions like fog and rain lead to image corruption which can cause a drastic drop in object detection (OD) performance. The primary navigation of autonomous vehicles depends on the effectiveness of the image processing techniques applied to the data collected from various visual sensors. Therefore, it is essential to develop the capability to detect objects like vehicles and pedestrians under challenging conditions such as like unpleasant weather. Ensembling multiple baseline deep learning models under different voting strategies for object detection and utilizing data augmentation to boost the models' performance is proposed to solve this problem. The data augmentation technique is particularly useful and works with limited training data for OD applications. Furthermore, using the baseline models significantly speeds up the OD process as compared to the custom models due to transfer learning. Therefore, the ensembling approach can be highly effective in resource-constrained devices deployed for autonomous vehicles in uncertain weather conditions. The applied techniques demonstrated an increase in accuracy over the baseline models and were able to identify objects from the images captured in the adverse foggy and rainy weather conditions. The applied techniques demonstrated an increase in accuracy over the baseline models and reached 32.75% mean average precision (mAP) and 52.56% average precision (AP) in detecting cars in the adverse fog and rain weather conditions present in the dataset. The effectiveness of multiple voting strategies for bounding box predictions on the dataset is also demonstrated. These strategies help increase the explainability of object detection in autonomous systems and improve the performance of the ensemble techniques over the baseline models.

Refining image‐velocimetry performances for streamflow monitoring: Seeding metrics to errors minimization

AbstractRiver streamflow monitoring is currently facing a transformation due to the emerging of new innovative technologies. Fixed and mobile measuring systems are capable of quantifying surface flow velocities and discharges, relying on video acquisitions. This camera‐gauging framework is sensitive to what the camera can “observe” but also to field circumstances such as challenging weather conditions, river background transparency, transiting seeding characteristics, among others. This short communication paper introduces the novel idea of optimizing image velocimetry techniques selecting the most informative sequence of frames within the available video. The selection of the optimal frame window is based on two reasonable criteria: (a) the maximization of the number of frames, subject to (b) the minimization of the recently introduced dimensionless seeding distribution index (SDI). SDI combines seeding characteristics such as seeding density and spatial clustering of tracers, which are used as a proxy to enhance the reliability of image velocimetry techniques. Two field case studies were considered as a proof‐of‐concept of the proposed framework, on which seeding metrics were estimated and averaged in time to select the proper application window. The selected frames were analysed using LSPIV to estimate the surface flow velocities and river discharge. Results highlighted that the proposed framework might lead to a significant error reduction. In particular, the computed discharge errors, at the optimal portion of the footage, were about 0.40% and 0.12% for each case study, respectively. These values were lower than those obtained, considering all frames available.

Cumbersome but desirable-Breaking the code of everyday cycling.

IntroductionCycling for transport could integrate physical activity (PA) into daily routines and potentially increase total PA levels. However, for parents with young children, most factors affecting transport mode choice tend to facilitate car use. Greater insight is necessary into reasons for (not) using sustainable transport modes in parents with young children. Therefore, the objective of this study was to explore the experiences, including motives, perceptions, attitudes, and norms, of parents of young children by using an e-bike, a longtail bike, and a traditional bike for everyday travel to the workplace, kindergarten, and the grocery store during the autumn, winter, and spring, in nine months.MethodsSemistructured focus group interviews were conducted with 18 parents of young children residing in southern Norway. Parents were recruited through Facebook announcements and direct contact with kindergartens, selected organisations, and companies in the Kristiansand municipality. Data were analysed by systematic text condensation by using NVivo V.11.ResultsParticipants’ experiences were summarised by three main themes: ‘cycling is cumbersome’, ‘cycling reflects the desirable me’, and ‘breaking the cycling code’. Time use, planning, logistics, wet and cold weather, long distances, and no cycling habit were frequently mentioned barriers, and the most notable facilitator was the children’s attitude towards cycling. In general, children loved to cycle and preferred cycling to driving. Additionally, the freedom and independence of cycling were emphasised and valued.ConclusionIn challenging weather conditions, parents of young children may experience cycling as cumbersome but desirable, and bike access could increase the feasibility of daily cycling.

Advances in Experimental Vehicle Soiling Tests

<div class="section abstract"><div class="htmlview paragraph">The field of vision of the driver during wet road conditions is essential for safety at all times. Additionally, the safe use of the increasing number of sensors integrated in modern cars for autonomous driving and intelligent driver assistant systems has to be ensured even under challenging weather conditions. To fulfil these requirements during the development process of new cars, experimental and numerical investigations of vehicle soiling are performed. This paper presents the surface contamination of self- and foreign-soiling tested in the wind tunnel. For these type of tests, the fluorescence method is state-of-the-art and widely used for visualizing critical areas. In the last years, the importance of parameters like the contact angle have been identified when designing the experimental setup. In addition, new visualization techniques have been introduced. Based on these developments the complete process using the fluorescence method was evaluated and updated to the available technology. Various factors in the experimental setup were identified to improve the reliability and a resulting high repeatability of the experiments between different test sessions is shown. Additionally, new visualization techniques are introduced for enhanced data evaluation which allows an easier identification of possible measures. Due to the increased repeatability, it is now possible to calculate the absolute difference between two analyzed vehicle configurations. For the side mirror glass, the quality of the results is improved by image detection methods. In the end it is shown that these modifications to the process are helpful in evaluating the results of computational fluid dynamics.</div></div>

Autonomous Navigation in Inclement Weather Based on a Localizing Ground Penetrating Radar

Most autonomous driving solutions require some method of localization within their environment. Typically, onboard sensors are used to localize the vehicle precisely in a previously recorded map. However, these solutions are sensitive to ambient lighting conditions such as darkness and inclement weather. Additionally, the maps can become outdated in a rapidly changing environment and require continuous updating. While LiDAR systems don't require visible light, they are sensitive to weather such as fog or snow, which can interfere with localization. In this letter, we utilize a Ground Penetrating Radar (GPR) to obtain precise vehicle localization. By mapping and localizing using features beneath the ground, we obtain features that are both stable over time, and maintain their appearance during changing ambient weather and lighting conditions. We incorporate this solution into a full-scale autonomous vehicle and evaluate the performance on over 17 km of testing data in a variety of challenging weather conditions. We find that this novel sensing modality is capable of providing precise localization for autonomous navigation without using cameras or LiDAR sensors.

LARUS: an unmanned aircraft for the support of maritime rescue missions under heavy weather conditions

In maritime emergencies, the time until arrival at the scene of an accident is a decisive factor for a successful rescue operation. In this context, the research project LARUS aims to support the rescue forces at sea with unmanned aerial systems and thus to optimize the rescue process. This refers on the one hand to the localization of potential accident sites and on the other hand to informing the rescuers about the situation on site as well as the provision of an efficient communication infrastructure. For this purpose, the UAS is equipped with various sensors and communication media. To ensure a seamless rescue procedure, the aircraft must be capable of operating even under severe weather conditions. In this regard, the Institute of Flight System Dynamics is currently investigating measures to reduce the gust loads to make the UAS robust against challenging weather conditions. The integration of the LARUS aerial system into airspace using an active collision avoidance system is also an important research aspect. In addition, the adaptation of flight control to the expanded requirements and influencing factors as well as the implementation of international SAR search strategies for flight guidance are also the subject of development work.

Automatic Vehicle Detection With Roadside LiDAR Data Under Rainy and Snowy Conditions

The previous studies showed that rainy and snowy weather can reduce the quality of LiDAR data. In rainy and snowy weather, laser beams of LiDAR were often blocked by raindrops or snowflakes, which was called weather occlusion. The vehicle detection with weather occlusion is a challenge. When the traditional density-based spatial clustering of applications with noise (DBSCAN) was used for vehicle clustering, the data processing showed that the false detection rate of the conventional DBSCAN under the snowy weather was high. This paper aims to present the characteristics of roadside LiDAR data in snowy and rainy days and improve the accuracy of vehicle detection during challenging weather conditions. A revised DBSCAN method named 3D-SDBSCAN is raised up to distinguish vehicle points and snowflakes in the LiDAR data. Adaptive parameters were applied in the revised DBSCAN method to detect vehicles with different distances from the roadside LiDAR sensor. The performance of the proposed method and the conventional DBSCAN algorithm were compared using the data collected under rainy and snowy conditions. The results showed that the 3D-SDBSCAN algorithm could overcome weather occlusion issue better than the conventional one.

The most difficult at-fault fatal crashes to avoid with current active safety technology

ObjectiveWe studied which current fatal at-fault crashes would occur despite the most advanced current active safety devices (up to SAE level 2 of driving automation) and how frequent these crashes would be. MethodsWe carried out a cross-sectional study of passenger cars that were first registered during the period 1st January 2010 to 31st December 2017 in Finland. To gain the true exposure for these cars, we accessed the national Vehicular and Driver Data Register to obtain the mileage information and the registration count for the study period of 2010-17. Similarly, we accessed the registry of Finnish road accident investigation teams and included all fatal at-fault crashes among the cars in our study for the same period. We used a real world reference technology for each active safety system in our analysis and chose one car brand as an example. This gave us exact system specifications and enabled testing the operation of the systems on the road. We performed field tests to gain further information on the precise operation of the safety systems in different operating conditions. Finally, we gathered all information on the studied active safety systems and analyzed the investigated at-fault fatal crashes case-by-case using our four level method. ResultsCars in our study were the primary party in 113 investigated fatal accidents during the years 2010-17. In 87 of the accidents, the leading cause of death was the injuries due to the crash, and these cases were classified as “unavoidable” (n = 58, 67 %), “avoidable” (n = 26, 30 %) or unsolved (n = 3, 3 %). Of the 58 “unavoidable” crashes 21 (36 %) were suicides, 21 (36%) involved active driver input which would have prevented the safety system operation, 15 (17 %) featured circumstances beyond the safety system performance and in one loss-of-control crash the driver had disabled the relevant safety system (electronic stability control). The registration years of the cars in our study (2010-17) totaled 3,772,864 and during this period, the cars travelled 75.9 billion kilometers. The crash incidence of the “unavoidable” at-fault fatal crashes was 0.76-0.80 fatal crashes per billion kilometers and 15-16 fatal crashes per million registration years. ConclusionsWe calculated a crash incidence for the “unavoidable” crashes which was 20–27% smaller than the observed crash rate of ESC-fitted passenger cars in our previous study. We concluded that suicides, active driver input until the crash, and challenging weather and road conditions are the most difficult factors for current active safety systems. Our analysis did not account for issues such as system usability or driver acceptance and therefore our results should be regarded as something that is currently theoretically achievable. However, the observed incidence is a good reference for automated driving development and the crash rate of automated cars.

Life cycle assessment of winter road maintenance

Winter road maintenance in the Nordic climate is demanding due to challenging weather conditions, high precipitation, and icy conditions. As a leading country in the transition to low-emission transport, Norway must work to reduce their emissions while providing a safe level of service through winter maintenance operations. This article investigates the environmental impacts of winter road maintenance (WRM) in Norway both today and under a climate change scenario predicted for 2050. Life cycle assessment (LCA) is used to evaluate the environmental impact of the functional unit “average winter road maintenance in Norway on national and county roads per km.lane.” The ReCiPe (hierarchy) method was used to identify and categorize emissions related to WRM to show how different factors affect the system and to reveal hidden emissions hotspots. Real-time data from WRM vehicles were used to determine how fuel consumption is affected by gradient and weather. Producers and operators provided other relevant information on WRM vehicles. Official reports supplied information on deicer quantities used and the total distance driven by WRM vehicles in Norway. The quantity of deicer used is the main source of emissions contributing toward all impact categories. The effect of deicer is likely to be even higher in certain impact categories. The environmental impact of the deicer after application is not included. The representation of WRM in existing emissions data is limited despite the considerable amount of deicer applied and the long distances that WRM vehicles travel. The results document how energy use throughout the system is another important source of emissions. Various parameters, such as road gradient, vehicle properties, driver behavior, and weather, affect the fuel consumption of WRM vehicles, with weather being the most important of these. Significant potential for emissions reductions from WRM was found, and WRM operations should be included in cold-climate road LCA studies. The environmental impacts of deicer application are especially high compared to the mechanical clearing of roads and contribute strongly to impact categories such as terrestrial, freshwater, and human toxicity and to the formation of particulate matter.

Effects of Ground Manifold Modeling on the Accuracy of Stixel Calculations

This paper highlights the role of ground manifold modeling for stixel calculations; stixels are medium-level data representations used for the development of computer vision modules for self-driving cars. By using single-disparity maps and simplifying ground manifold models, calculated stixels may suffer from noise, inconsistency, and false-detection rates for obstacles, especially in challenging datasets. Stixel calculations can be improved with respect to accuracy and robustness by using more adaptive ground manifold approximations. A comparative study of stixel results, obtained for different ground-manifold models (e.g., plane-fitting, line-fitting in $v$ -disparities or polynomial approximation, and graph cut), defines the main part of this paper. This paper also considers the use of trinocular stereo vision and shows that this provides options to enhance stixel results, compared with the binocular recording. Comprehensive experiments are performed on two publicly available challenging datasets. We also use a novel way for comparing calculated stixels with ground truth. We compare depth information, as given by extracted stixels, with ground-truth depth, provided by depth measurements using a highly accurate LiDAR range sensor (as available in one of the public datasets). We evaluate the accuracy of four different ground-manifold methods. The experimental results also include quantitative evaluations of the tradeoff between accuracy and run time. As a result, the proposed trinocular recording together with graph-cut estimation of ground manifolds appears to be a recommended way, also considering challenging weather and lighting conditions.

Deterioration Characteristics of Waterborne Pavement Markings Subjected to Different Operating Conditions

AbstractDrivers rely on pavement markings to maintain a safe road path, especially during nighttime and challenging weather conditions. The objective of this study was to model the deterioration of...

Deep learning–assisted real-time container corner casting recognition

Intelligent automated crane systems are now an integral part of container port automation. Accurate corner casting detection boosts the performance of an automated crane system which ultimately automates ships loading and unloading. Existing techniques use various traditional laser-based and vision-based methods for corner casting detection. Challenging weather conditions, varying lighting conditions, light reflections from ground, and container rusting conditions are the main problems that affect the performance of automated cranes. From this line of research, we propose an end-to-end method that takes a low-quality video input and produces bounding boxes around corner castings by applying a recurrent neural network along with long short-term memory units. The expressive image features from GoogLeNet are used to produce intermediate image representations that are further tuned for our system. The proposed system uses back-propagation to allow joint tuning of all components. At least, four cameras are mounted on each crane and input stream is combined into a single image to reduce the computational cost. The proposed system outperforms all existing methods in terms of precision, recall, and F-measure. The proposed method is implemented in a real-time port and produces more than 98% accuracy in all conditions.