The aims of the study were to analyze metabolic power (MP) and MP derived parameters for different positions in the German Soccer Bundesliga and to evaluate if classification of high-intensity is more suited using the metabolic power approach instead of using traditional speed-based methods. 1,345 video match analysis (25 Hz) datasets from 380 players in 96 matches of the German first Bundesliga were gathered by an automatic player tracking system. Displacement (speed, acceleration, distance) and energetic (MP, energy expenditure) variables were determined. Intensity was classified utilizing conventional thresholds. Metabolic and running profiles were compared among six positional groups and between the halves of the match respectively (one-way ANOVA). Further, time spent, distance covered and energy expended at high speed (>15.5 km h −1 ) and high acceleration (>2 m s −2 ) were compared to those at high MP (>20 W kg −1 ) (one-way ANOVA) for evaluating if metabolic power is more suited to describe intensity in team sports. Main findings are that central-attacking midfielders (CAM) and central midfielders (CM) expended more energy (CAM: 59.8 ± 4.2 kJ kg −1 ; CM: 59.6 ± 3.6 kJ kg −1 ) and covered more distance (CAM: 11,494 ± 765 m; CM: 11,445 ± 638 m) than all other positions ( p < 0.001). In the whole group of players, less time (t) and less energy were expended (EE) in high speed (t: 302 ± 84 s; EE: 10.1 ± 2.9 kJ kg −1 ) and at high acceleration zones (t: 147 ± 24 s; EE: 5.5 ± 1.2 kJ kg −1 ) when compared to high MP zone (t: 617 ± 141 s; EE: 20.0 ± 4.4 kJ kg −1 ) ( p < 0.001). Furthermore, players covered more distance at high MP (2,464 ± 597 m) than at high speed (1,638 ± 458 m) and much less at high acceleration (359 ± 67 m) ( p < 0.001). The higher activity profiles of CAM and CM compared to the other positions indicate the need for higher developed physiological performance in players of these positions. High intensity activities should be interpreted differently when using MP and displacement parameters as indicators of high intensity in soccer.

Computer-aided sports analysis is demanded by coaches and the media. Image processing and machine learning techniques that allow for "live" recognition and tracking of players exist. But these methods are far from collecting and analyzing event data fully autonomously. To generate accurate results, human interaction is required at different stages including system setup, calibration, supervision of classifier training, and resolution of tracking conflicts. Furthermore, the real-time constraints are challenging: in contrast to other object recognition and tracking applications, we cannot treat data collection, annotation, and learning as an offline task. A semi-automatic labeling of training data and robust learning given few examples from unbalanced classes are required. We present a real-time system acquiring and analyzing video sequences from soccer matches. It estimates each player's position throughout the whole match in real-time. Performance measures derived from these raw data allow for an objective evaluation of physical and tactical profiles of teams and individuals. The need for precise object recognition, the restricted working environment, and the technical limitations of a mobile setup are taken into account. Our contribution is twofold: (1) the deliberate use of machine learning and pattern recognition techniques allows us to achieve high classification accuracy in varying environments. We systematically evaluate combinations of image features and learning machines in the given online scenario. Switching between classifiers depending on the amount of training data and available training time improves robustness and efficiency. (2) A proper human-machine interface decreases the number of required operators who are incorporated into the system's learning process. Their main task reduces to the identification of players in uncertain situations. Our experiments showed high performance in the classification task achieving an average error rate of 3 % on three real-world datasets. The system was proved to collect accurate tracking statistics throughout different soccer matches in real-time by incorporating two human operators only. We finally show how the resulting data can be used instantly for consumer applications and discuss further development in the context of behavior analysis.

While much work in the domain of traffic lights recognition is invested in the detection of traffic lights, classification of their exact state (including color phase and possible arrow pictogram) is often neglected. In this paper, we propose a robust approach for efficient video-based classification of said state with particular attention to the displayed pictogram and an additional ability to reject false detections. The currently active lights are identified and used to classify the phase. The lights are extracted and transformed into a HOG feature representation that is used to classify the pictogram with the help of machine learning classifiers. In order to gain optimal results, we compared the performance of different algorithms, namely LDA, kNN, and SVM. We provide an evaluation of our method on individual images and demonstrate that the classification rate of the phase lies at 96.7% and at 92.8% for the pictogram, with the use of SVMs providing best results. This leads to an overall classification quality of 89.9%. With a runtime of less than 1ms per image section our algorithm can easily be integrated in every traffic light recognition pipeline.

The time-consuming search for parking lots could be assisted by efficient routing systems. Still, the needed vacancy detection is either very hardware expensive, lacks detail or does not scale well for industrial application. This paper presents a video-based system for cost-effective detection of vacant parking lots, and an extensive evaluation with respect to the system's transferability to unseen environments. Therefore, different image features and learning algorithms were examined on three independent datasets for an unbiased validation. A feature / classifier combination which solved the given task against the background of a robustly scalable system, which does not require re-training on new parking areas, was found. In addition, the best feature provides high performance on gray value surveillance cameras. The final system reached an accuracy of 92.33% to 99.96%, depending on the parking rows' distance, using DoG-features and a support vector machine.

In this study, we present a new indoor positioning and environment perception system for generic objects based on multiple surveillance cameras. In order to assist highly automated driving, our system detects the vehicle's position and any object along its current path to avoid collisions. A main advantage of the proposed approach is the usage of cameras that are already installed in the majority of parking garages. We generate precise object hypotheses in 3D world coordinates based on a given extrinsic camera calibration. Starting with a background subtraction algorithm for the segmentation of each camera image, we propose a robust view-ray intersection approach that enables the system to match and triangulate segmented hypotheses from all cameras. Comparing with LIDAR-based ground truth, we were able to evaluate the system's mean localization accuracy of 0.37 m for a variety of different sequences.

Semi-Global Matching (SGM) is arguably one of the most popular algorithms for real-time stereo vision. It is already employed in mass production vehicles today. Thinking of applications in intelligent vehicles (and fully autonomous vehicles in the long term), we aim at further improving SGM regarding its accuracy. In this study, we propose a straight-forward extension of the algorithm's parametrization. We consider individual penalties for different path orientations, weighted integration of paths, and penalties depending on intensity gradients. In order to tune all parameters, we applied evolutionary optimization. For a more efficient offline optimization and evaluation, we implemented SGM on graphics hardware. We describe the implementation using CUDA in detail. For our experiments, we consider two publicly available datasets: the popular Middlebury benchmark as well as a synthetic sequence from the .enpeda. project. The proposed extensions significantly improve the performance of SGM. The number of incorrect disparities was reduced by up to 27.5 % compared to the original approach, while the runtime was not increased.

Real-time detection of traffic signs, the task of pinpointing a traffic sign's location in natural images, is a challenging computer vision task of high industrial relevance. Various algorithms have been proposed, and advanced driver assistance systems supporting detection and recognition of traffic signs have reached the market. Despite the many competing approaches, there is no clear consensus on what the state-of-the-art in this field is. This can be accounted to the lack of comprehensive, unbiased comparisons of those methods. We aim at closing this gap by the “German Traffic Sign Detection Benchmark” presented as a competition at IJCNN 2013 (International Joint Conference on Neural Networks). We introduce a real-world benchmark data set for traffic sign detection together with carefully chosen evaluation metrics, baseline results, and a web-interface for comparing approaches. In our evaluation, we separate sign detection from classification, but still measure the performance on relevant categories of signs to allow for benchmarking specialized solutions. The considered baseline algorithms represent some of the most popular detection approaches such as the Viola-Jones detector based on Haar features and a linear classifier relying on HOG descriptors. Further, a recently proposed problem-specific algorithm exploiting shape and color in a model-based Houghlike voting scheme is evaluated. Finally, we present the best-performing algorithms of the IJCNN competition.

The search for free parking space in a crowded car park is a time-consuming and tedious task. Today's park assistance systems provide the driver with acoustic or visual feedback when approaching an obstacle or semi-autonomously navigate the vehicle into the parking lot. However, finding a free parking lot is usually left to the driver. In this paper, we address this search problem via video sensors only. This can be used as a help to the driver to quickly pass a parking deck and, more important, can be regarded as a cornerstone to fully autonomously parking vehicles.

Finding a vacant parking lot in urban areas is mostly time-consuming and not satisfying for potential visitors or customers. Efficient car-park routing systems could support drivers to find a nun occupied parking lot. Current systems detecting vacant parking lots are either very expensive due to the hardware requirement or do not provide a detailed occupancy map. In this paper, we propose a video-based system for low-cost parking space classification. A wide-angle lens camera is used in combination with a desktop computer. We evaluate image features and machine learning algorithms to determine the occupancy of parking lots. Each combination of feature set and classifier was trained and tested on our dataset containing approximately 10,000 samples. We assessed the performance of all combinations of feature extraction and classification methods. Our final system, incorporating temporal filtering, reached an accuracy of 99.8 %.

Die Automatisierung der Videoanalyse nimmt im Profisport eine immer wichtigere Rolle ein. Im Fußball kommt dabei der Auswertung der Laufwege der Spieler eine besondere Bedeutung zu. Der vorliegende Bericht dokumentiert unser Kooperationsprojekt zum computergestützten Spieler-Tracking auf Basis von Videobildern in Echtzeit. Wir beschreiben den Aufbau und diskutieren die Praxistauglichkeit des entwickelten Systems, das sich durch hohe Genauigkeit, Mobilität und Kostengünstigkeit auszeichnet.

In this paper, we propose a new approach for localization and tracking of a vehicle in a parking garage, based on environment-embedded LIDAR sensors. In particular, we present an integration of data from multiple sensors, allowing to track vehicles in a common, parking garage coordinate system. In order to perform detection and tracking in realtime, a combination of appropriate methods, namely a grid-based approach, a RANSAC algorithm, and a Kalman filter is proposed and evaluated. The system achieves highly confident and exact vehicle positioning. In the context of a larger framework, our approach was used as a reference system to enable autonomous driving within a parking garage. In our experiments, we showed that the proposed algorithm allows a precise vehicle localization and tracking. Our system's results were compared to human-labeled ground-truth data. Based on this comparison we prove a high accuracy with a mean lateral and longitudinal error of 6.3cm and 8.5 cm, respectively.

For the development of vision-based driver assistance systems, large amounts of data are needed, e.g., for training machine learning approaches, tuning parameters, and comparing different methods. There are basically three possible ways to obtain the required data: using freely available benchmark sets, doing own recordings, or falling back to synthesized sequences. In this paper, we show that Google Street View can be incorporated as a valuable source for image data. Street View is the largest publicly available collection of images recorded from a drivers' perspective, covering many different countries and scenarios. We describe how to efficiently access the data and present a framework that allows for virtual driving through a network of images. We assess its performance and show its applicability in practice considering traffic sign recognition as an example. The introduced approach supports an efficient collection of image data relevant to training and evaluating machine vision modules. It is easily adaptable and extendible, whereby Street View becomes a valuable tool for developers of vision-based assistance systems.

Traffic signs are characterized by a wide variability in their visual appearance in real-world environments. For example, changes of illumination, varying weather conditions and partial occlusions impact the perception of road signs. In practice, a large number of different sign classes needs to be recognized with very high accuracy. Traffic signs have been designed to be easily readable for humans, who perform very well at this task. For computer systems, however, classifying traffic signs still seems to pose a challenging pattern recognition problem. Both image processing and machine learning algorithms are continuously refined to improve on this task. But little systematic comparison of such systems exist. What is the status quo? Do today's algorithms reach human performance? For assessing the performance of state-of-the-art machine learning algorithms, we present a publicly available traffic sign dataset with more than 50,000 images of German road signs in 43 classes. The data was considered in the second stage of the German Traffic Sign Recognition Benchmark held at IJCNN 2011. The results of this competition are reported and the best-performing algorithms are briefly described. Convolutional neural networks (CNNs) showed particularly high classification accuracies in the competition. We measured the performance of human subjects on the same data-and the CNNs outperformed the human test persons.

Neue Fahrerassistenzsysteme versprechen ein verbessertes Fahrverhalten und einen zusaetzlichen Sicherheitsgewinn fuer Motorraeder. Dabei ist die Bestimmung des Rollwinkels ein wichtiger Bestandteil. Videobasierte Assistenzfunktionen sind in anderen Fahrzeugklassen bereits etabliert. Deren Uebertragung auf Motorraeder ist jedoch mit grossen Herausforderungen verbunden, da die Schraeglage grosse Variationen in den aufgenommenen Bildern bewirkt. Um diesen Einfluss zu kompensieren und videobasierte Fahrerassistenzsysteme fuer Motorraeder zu realisieren, ist eine genaue Bestimmung des Rollwinkels erforderlich.In dieser Studie werden daher ein auf Fahrdynamiksensorik und ein auf Videosensorik basierender Ansatz zur Bestimmung des Rollwinkels vorgestellt, verglichen und bewertet. Fuer den Fahrdynamik-Ansatz wird ein Extended-Kalman-Filter eingesetzt, der auf Basis von Modellen zur Rolldynamik und der Inertialsensorik den Rollwinkel des Motorrades schaetzt. Der Video-Ansatz nutzt sich wiederholende Geometrie- und Objektzusammensetzungen der aufgenommenen Szenen aus. Durch das Aufzeichnen von Fahrsituationen mit bekanntem Rollwinkel lassen sich fuer jede Schraeglage charakteristische Orientierungshistogramme der Bildgradienten finden. Durch Korrelation einer aktuellen Aufnahme mit der gelernten Verteilung kann der Rollwinkel abgeleitet werden. Die Evaluierung der beiden beschriebenen Ansaetze basiert auf Manoevern, die relevante Fahrdynamikbereiche abdecken. Ergaenzt wird dies durch Fahrten auf einer Strecke, die hinsichtlich der Ortslage das deutsche Strassennetz repraesentativ abbildet. Ein Vergleich der ermittelten Rollwinkelverlaeufe mit einer hochgenauen Referenzsensorik zeigt, dass beide Verfahren hohe Genauigkeiten im Bereich von circa 2 Grad erzielen, wobei sich wie erwartet Unterschiede zeigen, je nachdem welche Situation vorliegt. Eine Kombination beider Verfahren bietet somit Potenzial fuer viele weitere videobasierte Anwendungen im Motorrad. ABSTRACT IN ENGLISH: New Advanced Rider Assistance Systems (ARAS) for Powered Two-Wheelers (PTW) promise improved comfort and safety, based on roll angle estimation as a key element. Video-based assistance systems are already successfully deployed in cars. Their portation to PTWs requires information about the camera pose, since large roll angles produce significant variations in the recorded images. Therefore, accurate roll angle estimation is essential for video-based ARAS. In this paper two approaches for roll angle estimation are introduced, compared and evaluated. One is based on vehicle dynamics sensors, the other one on video data. For the vehicle dynamics approach, an extended Kalman filter is used. This filter estimates the PTW’s roll angle by deploying models for roll dynamics and sensor characteristics. The video-based approach makes use of recurring geometry and object composition of the recorded scenes. By acquiring video data for maneuvers with given roll angle, characteristic orientation histograms of the image gradient can be found. Correlating new recordings with the learnt statistics, the current roll angle can be derived. For evaluation, riding maneuvers covering all relevant vehicle dynamics are conducted. Furthermore, rides on a public road course representative for the German road network are carried out. The estimated roll angles of both approaches are compared with a highly precise reference sensor. A high accuracy of about 2 degrees error can be observed for both methods. A combination of both indicates a high potential for many new video-based applications in PTWs. (A)

Advanced Rider Assistance Systems (ARAS) for powered two-wheelers improve driving behaviour and safety. Further developments of intelligent vehicles will also include video-based systems, which are successfully deployed in cars. Porting such modules to motorcycles, the camera pose has to be taken into account, as e. g. large roll angles produce significant variations in the recorded images. Therefore, roll angle estimation is an important task for the development of various kinds of ARAS. This study introduces alternative approaches based on inertial measurement units (IMU) as well as video only. The latter learns orientation distributions of image gradients that code the current roll angle. Until now only preliminary results on synthetic data have been published. Here, an evaluation on real video data will be presented along with three valuable improvements and an extensive parameter optimisation using the Covariance Matrix Adaptation Evolution Strategy. For comparison of the very dissimilar approaches a test vehicle is equipped with IMU, camera and a highly accurate reference sensor. The results state high performance of about 2 degrees error for the improved vision method and, therefore proofs the proposed concept on real-world data. The IMU-based Kalman filter estimation performed on par. As a naive result averaging of both estimates already increased performance an elaborate fusion of the proposed methods is expected to yield further improvements.

Traffic signs are characterized by a wide variability in their visual appearance in real-world environments. For example, changes of illumination, varying weather conditions and partial occlusions impact the perception of road signs. In practice, a large number of different sign classes needs to be recognized with very high accuracy. Traffic signs have been designed to be easily readable for humans, who perform very well at this task. For computer systems, however, classifying traffic signs still seems to pose a challenging pattern recognition problem. Both image processing and machine learning algorithms are continuously refined to improve on this task. But little systematic comparison of such systems exist. What is the status quo? Do today's algorithms reach human performance? For assessing the performance of state-of-the-art machine learning algorithms, we present a publicly available traffic sign dataset with more than 50,000 images of German road signs in 43 classes. The data was considered in the second stage of the German Traffic Sign Recognition Benchmark held at IJCNN 2011. The results of this competition are reported and the best-performing algorithms are briefly described. Convolutional neural networks (CNNs) showed particularly high classification accuracies in the competition. We measured the performance of human subjects on the same data-and the CNNs outperformed the human test persons.