We propose a framework for detecting and tracking multiple interacting objects from a single, static, uncalibrated camera. The number of objects is variable and unknown, and object-class-specific models are not available. We use background subtraction results as measurements for object detection and tracking. Given these constraints, the main challenge is to associate pixel measurements with (possibly interacting) object targets. We first track clusters of pixels, and note when they merge or split. We then build an inference graph, representing relations between the tracked clusters. Using this graph and a generic object model based on spatial connectedness and coherent motion, we label the tracked clusters as whole objects, fragments of objects or groups of interacting objects. The outputs of our algorithm are entire tracks of objects, which may include corresponding tracks from groups of objects during interactions. Experimental results on multiple video sequences are shown.

Τα τελευταία χρόνια η ενεργειακή κρίση έχει αρχίσει και γίνεται ορατή σε όλες τις εκφάνσεις της καθημερινής ζωής, επηρεάζοντάς την σε σημαντικό βαθμό. Τα αποθέματα πετρελαίου λιγοστεύουν με ταχύτατους ρυθμούς ενώ η ζήτηση αυξάνεται με φρενήρεις ρυθμούς. Ταυτόχρονα, οι φυσικές καταστροφές, σημαντικό μέρος των οποίων αποδίδεται στην κλιματική αλλαγή, που με τη σειρά της αποδίδεται –εν μέρει- στην ρύπναση εξαιτίας της παραγωγής και χρήσης συμβατικών μορφών ενέργειας, και στη συνεχιζόμενη μόλυνση του περιβάλλοντος κατέστησαν αναγκαία την εύρεση και εφαρμογή εναλλακτικών μορφών παραγωγής ηλεκτρικής ενέργειας, φιλικές προς αυτό. Λύση στο αδιέξοδο αυτό φαίνεται να δίνουν οι Ανανεώσιμες Πηγές Ενέργειας καθώς σιγά-σιγά κατακτούν ολοένα και περισσότερο έδαφος στο χώρο της ενέργειας. Μορφές Ανανεώσιμων Πηγών Ενέργειας αποτελούν η αιολική ενέργεια, η ηλιακή, η υδροηλεκτρική, η ενέργεια από βιομάζα, η γεωθερμία, η παλιρροιακή ενέργεια, κ.α. οι οποίες αναπτύσσονται.Η παρούσα διπλωματική εργασία ασχολείται με την μελέτη της χρησης μιας καινοτόμας τεχνολογιας των φωτοβολταικών μονάδων με το όνομα ηλιακοί ιχνηλάτες.Αρχικά πραγματοποιείται εκτενής αναφορά στην βασική θεωρία της φωτοβολταικής τεχνολογίας δηλαδή στο φωτοβολταικό φαινόμενο,στα φωτοβολταικά κύταρρα και στην μοντελοποίηση του φωτοβολταικού κυτάρρου με την βοήθεια του προγραμματιστικού περιβάλοντος ΜΑTLAB.Στην συνέχεια παρουσιάζονται βασικές γνώσεις πάνω στην ηλιακή ακτινοβολία και την θέση του ηλίου σε σχέση με την γη που ειναι απαραίτητες αφου σε αυτες στηρίζονται οι βασικοί αστρονομικοί αλγόριθμοι που χρησιμοποιούν τα συστήματα ιχνηλάτισης του ηλίου που μελετώνται.Ακολουθεί εκτενής αναφορά στην θεωρία των συστημάτων ιχνηλάτισης ανοιχτού και κλειστού βρόχου καθως και στους σημαντικότερους αστρονομικούς αλγορίθμους που εχουν επινοηθεί μεχρι σήμερα.Επίσης γίνεται συγκριτική μελέτη με την βοήθεια προσομοιώσεων του ΜΑΤLAB μεταξύ των σταθερών πάνελ και αυτών με ιχνηλάτιση ανοιχτού βρόχου.Επίσης πραγματοποιείται η διάκριση μεταξύ μονοαξονικών και διαξονικών ηλιακών ιχνηλατών ανοιχτού και κλειστού βρόχου.Στο κεφάλαιο 5 παρατίθενται οι 2 αλγόριθμοι που κατασκευάσθηκαν ετσι ωστε να προσομοιώσουμε μέσω του ΜΑΤLAB την θέση του ηλίου κατα την διάρκεια μιας ολόκληρης ημέρας για οποιαδήποτε χρονολογία και τοποθεσία επυθημούμε καθως και για να υπολογίσουμε την ολική ευθεία, διάχυτη και ανακλώμενη ηλιακή ακτινοβολία εκείνης της ημέρας!Τρέχοντας λοιπόν μια ολοήμερη προσομοίωση λαμβάνουμε τα απαραίτητα στοιχεία(μετρήσεις) ωστε να να πραγματοποιήσουμε μια συγκριτική μελέτη μεταξύ των συστημάτων ιχνηλάτισης ανοιχτού βρόχου που χρησιμοποιούν τον αστρονομικό αλγόριθμο που δημιουργήσαμε και των ηλιακών ιχνηλατών κλειστού βρόχου που με την βοήθεια πυρανομέτρου μετρούν την ηλιακή ακτινοβολία. Παράλληλα χρησιμοποιώντας τον αλγόριθμο που κατασκευάσαμε για προσομοίωση της τροχιάς του ηλίου δημιουργήσαμε ενα πρόγραμμα στο MATLAB με το οποιο εισάγοντας τις γεωγραφικές συντεταγμένες του τόπου που μας ενδιαφέρει αλλα και την παγκόσμια ημερομηνία και ώρα παίρνουμε μια τρισδιάστατη απεικόνιση της θέσης στην οποία πρέπει να τοποθετήσουμε το σταθερό φωτοβολταικό μας πλαίσιο εαν θέλουμε να επυτήχουμε την μέγιστη απόδοση.Αυτή είναι μια πολυ σημαντική λεπτομέρεια αφου ετσι πραγματοποιείται και σύγκριση ολων των προηγούμενων με τα σταθερά πάνελ που ομως εχουν την δυνατότητα εποχιακής ρύθμισης.Τελος προτείνεται η βέλτιστη οικονομικά αλλα και ενεργειακά λύση ωστε να επυτήχουμε την υψηλότερη απόδοση. Ολοι οι αλγόριθμοι που χρησιμοποιήθηκαν περιγράφονται και επεξηγούνται αναλυτικά σε κάθε κεφάλαιο. / This analysis deals with the study of the use of an innovative technology for photovoltaic modules named Solar trackers.Initially it is being made extensive referaince to the basic theory of photovoltaic technology as the photovoltaic effect, the photovoltaic cells simulated with the help of programming environment of MATLAB.Also the basic knowledge of solar radiation and the position of the sun in relathion to earth are being illustrated since these are based on the astronomical algorithms that tracking systems use.There is also an extensive raiference to open and closed loop tracking systems as well as the most important astronomical algorithms have been devised up to now.Also a comparative study with the help of MATLAB simulations between fixed panels and with an open loop tracking system has been made.Last we made the distinction between single-axle and two axles solar trackers open and closed loop.At Chapter 5 lists the 2 algorithms constructed so as to simulate with MATLAB the position of the sun during a whole day for any date and location we desire and to estimate the total direct, diffuse and reflected sunlight that day!

Made available in DSpace on 2014-06-11T19:22:52Z (GMT). No. of bitstreams: 0 Previous issue date: 2006-04-28Bitstream added on 2014-06-13T19:49:11Z : No. of bitstreams: 1 moura_fa_me_rcla.pdf: 671291 bytes, checksum: ceb37cf4906958931df2e8b048bf3133 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O objetivo deste estudo foi analisar as ações e seqüências de ações técnicas de jogadores e as estratégias de finalizações durante quatro partidas de futebol, a partir do tracking computacional. Foi utilizado o software Dvideow para a obtenção da posição dos jogadores em função do tempo e suas ações técnicas. De posse dos dados das ações dos jogadores, foram identificadas as ações e seqüências de ações realizadas e dados sobre as estratégias de finalizações utilizadas pelas equipes. Os resultados mostraram que a ação passe, e as seqüências de ações domínio?passe e domínio?condução?passe compreendem em média 89,5% das ações dos jogadores. Foram também as ações e seqüências de ações curtas as mais realizadas pelos jogadores no momento da finalização, além das cobranças de bola parada. Verificou-se que os atacantes foram os que mais realizaram seqüências de ações incompletas, enquanto os zagueiros foram os que menos realizaram. No entanto, os zagueiros foram os que mais desarmaram, enquanto que os atacantes foram os jogadores que mais finalizaram. Com relação às estratégias de finalizações, verificou-se que as equipes normalmente ganham a posse de bola em seu campo de defesa. Já o passe que antecede à finalização normalmente é executado no campo de ataque e possui até 20 metros. Os resultados também mostraram que as posses de bola que resultaram em finalizações envolvem poucos ou muitos passes e têm curta ou longa duração, dependendo da situação do jogo. Por fim, verificou-se que as finalizações ocorreram numa distância média de 21,02 metros. Assim, através desses resultados, conclui-se que jogadores normalmente realizam ações e seqüências de ações que envolvem poucos fundamentos e que as posses de bola que resultam em finalizações normalmente se iniciam no campo de defesa e podem envolver poucos ou muitos passes, e ter longa ou curta duração... (Resumo complet / The purpose of this study was to analyze the technical actions, the sequence of actions and the shots to goal strategies during four soccer matches, through computational tracking. Software Dvideow was used to obtain data about players position and their technical actions. Then, it was identified the technical actions and sequence of actions performed by players and the shots to goal strategies. The results showed that the pass movement, and the Control?Pass and Control? Dribbling? Pass sequence of actions corresponded to 89,5% of the players actions. Shorts actions and sequences of actions are the most frequent movements performed by players in shots to goal too, besides the free kicks and penalty kicks. It was verified that forwards were the players who most performed incomplete sequence of actions, while defenders were the players that performed the lower number of them. Nevertheless, the defenders were the players that most performed tackles, while the forwards were the players who most performed shots to goal. About the shots strategies, it was verified that the teams usually win the ball possession in their defensive soccer field. The pass executed before the shots to goal frequently occurred in attack soccer field and it had up to 20 meters of size. The results also showed that ball possessions that resulted in shots to goal can involve few or great number of passes, and they have short or long duration, depending on the game situation. Finally, it was found that shots to goal occurred on a 22.02 meters average distance. So, it was concluded that: ... (Complete abstract, click electronic access below)

This thesis focuses on inclusion of affiliate marketing to internet marketing as a contemporary modern tool for acquiring new customers in the online market and on analysis of the system solution for conversion tracking at eHUB.cz Ltd. The thesis is divided into three parts: the first part presents the general overview of affiliate marketing, the second section analyzes the current state of the conversion tracking system at eHUB.cz Ltd. and the third proposes an optimal solution for the future status of this system. Theoretical level corresponds to the first part and a practical second and third. The theoretical part clarifies the position of affiliate marketing within online marketing, including basic knowledge and relationships that are necessary for proceeding to the practical part. The practical part contains a description of the original and current technical solutions of system eHUB. At the same time suggests its optimization for a future state. The main objective is to find a precise definition of the current standards of conversion tracking system solutions in eHUB.cz Ltd. The main contribution of this work is the usability of the optimal technical solution at eHUB.cz Ltd.

In this thesis, we propose a new aerial video dataset and benchmark for low altitude UAV target tracking, as well as, a photo-realistic UAV simulator that can be coupled with tracking methods. Our benchmark provides the first evaluation of many state of-the-art and popular trackers on 123 new and fully annotated HD video sequences captured from a low-altitude aerial perspective. Among the compared trackers, we determine which ones are the most suitable for UAV tracking both in terms of tracking accuracy and run-time. We also present a simulator that can be used to evaluate tracking algorithms in real-time scenarios before they are deployed on a UAV ”in the field”, as well as, generate synthetic but photo-realistic tracking datasets with free ground truth annotations to easily extend existing real-world datasets. Both the benchmark and simulator will be made publicly available to the vision community to further research in the area of object tracking from UAVs. Additionally, we propose a persistent, robust and autonomous object tracking system for unmanned aerial vehicles (UAVs) called Persistent Aerial Tracking (PAT). A computer vision and control strategy is applied to a diverse set of moving objects (e.g. humans, animals, cars, boats, etc.) integrating multiple UAVs with a stabilized RGB camera. A novel strategy is employed to successfully track objects over a long period, by ’handing over the camera’ from one UAV to another. We integrate the complete system into an off- 4 the-shelf UAV, and obtain promising results showing the robustness of our solution in real-world aerial scenarios.

Relying on signs and symptoms of head injury outcomes has shown to be unreliable in capturing the vulnerabilities associated with brain trauma (Karton & Hoshizaki, 2018). To accommodate the subjectivity of self-reported symptoms, data collection using sensor monitoring and video analysis combined with event reconstruction are used to objectively measure trauma exposure (Tator, 2013; Scorza & Cole, 2019; Hoshizaki et al., 2014). Athletes are instrumented with wireless sensors designed to measure head kinematics during play. However, these systems have not been widely adopted as they are expensive, face challenges with angular acceleration measures, and often require video confirmation to remove false positives. Video analysis of head impacts, in conjunction with physical event reconstruction and finite element (FE) modeling, is also used to calculate tissue level strain. This data collection method requires specialized equipment and expertise. Effective management of head trauma in sport requires an objective, accessible, and quantifiable tool that addresses the limitations associated with current measurement systems. The purpose of this research was to determine if a simplified version of video analysis and event reconstruction using impact characteristics (velocity, location, mass, and compliance) obtained from body tracking could yield similar measures of brain strain magnitude to the standard head tracking method. Ice hockey impacts (x36) that varied in terms of competition level, event type and maximum principal strain (MPS) were chosen for analysis. 2D videos of previously completed head reconstructions were reanalyzed and each event was reconstructed again in the laboratory using impact parameters obtained from body tracking. MPS values were calculated using finite element (FE) modeling and compared to the MPS values from events that were reconstructed using impact parameters obtained from head tracking. The relationship between head and body tracking MPS data and level of agreement between MPS categories were also assessed. Overall, a significant difference was observed between MPS magnitudes obtained using impact parameters from body and head tracking data from 2D video. When analyzed by event type, only shoulder and glass events demonstrated significant differences in MPS magnitudes. A strong linear relationship was depicted between the two data collection methods and moderate level of agreement between MPS categories was observed, demonstrating that impact characteristics obtained from body tracking and 2D video can be used to measure brain tissue strain.

No description available.

No description available.

Multiple object tracking consists of detecting and identifying objects in video. In some applications, such as robotics and surveillance, it is desired that the tracking is performed in real-time. This poses a challenge in that it requires the algorithm to run as fast as the frame-rate of the video. Today's top performing tracking methods run at only a few frames per second, and can thus not be used in real-time. Further, when determining the speed of the tracker, it is common to not include the time it takes to detect objects. We argue that this way of measuring speed is not relevant for robotics or embedded systems, where the detecting of objects is done on the same machine as the tracking. We propose that one way of running a method in real-time is to not look at every frame, but skip frames to make the video have the same frame-rate as the tracking method. However, we believe that this will lead to decreased performance. In this project, we implement a multiple object tracker, following the tracking-by-detection paradigm, as an extension of an existing method. It works by modelling the movement of objects by solving the filtering problem, and associating detections with predicted new locations in new frames using the Hungarian algorithm. Three different similarity measures are used, which use the location and shape of the bounding boxes. Compared to other trackers on the MOTChallenge leaderboard, our method, referred to as C++SORT, is the fastest non-anonymous submission, while also achieving decent score on other metrics. By running our model on the Okutama-Action dataset, sampled at different frame-rates, we show that the performance is greatly reduced when running the model - including detecting objects - in real-time. In most metrics, the score is reduced by 50%, but in certain cases as much as 90%. We argue that this indicates that other, slower methods could not be used for tracking in real-time, but that more research is required specifically on this. / För att spåra rörliga objekt i video (eng: multiple object tracking) krävs att man lokaliserar och identifierar dem. I vissa tillämpningar, såsom robotik och övervakning, kan det krävas att detta görs i realtid, vilket kan vara svårt i praktiken, då det förutsätter att algoritmen kan köras lika fort som videons bildfrekvensen. De kraftfullaste algoritmerna idag kan bara analysera ett fåtal bildrutor per sekund, och lämpar sig därför inte för realtidsanvändning. Dessutom brukar tiden per bildruta inte inkludera den tid det tar att lokalisera objekt, när hastigheten av en algoritm presenteras. Vi anser att det sättet att beräkna hastigheten inte är lämpligt inom robotik eller inbyggda system, där lokaliseringen och identifiering av objekt sker på samma maskin. Många algoritmer kan köras i realtid genom att hoppa över det antal bildrutor i videon som krävs för att bildfrekvensen ska bli densamma som algoritmens frekvens. Dock tror vi att detta leder till sämre prestanda. I det här projektet implementerar vi en algoritm för att identifiera rörliga objekt. Vår algoritm bygger på befintliga metoder inom paradigmen tracking-by-detection (ung. spårning genom detektion). Algoritmen uppskattar hastigheten hos varje objekt genom att lösa ett filtreringsproblem. Utifrån hastigheten beräknas en förväntad ny position, som kopplas till nya observationer med hjälp av Kuhn–Munkres algoritm. Tre olika likhetsmått används, som på olika sätt kombinerar positionen för och formen på objekten. Vår metod, C++SORT, är den snabbaste icke-anonyma metoden publicerad på MOTChallenge. Samtidigt presterar den bra enligt flera andra mått. Genom att testa vår algoritm på video från Okutama-Action, med varierande bildfrekvens, kan vi visa att prestandan sjunker kraftigt när hela modellen - inklusive att lokalisera objekt - körs i realtid. Prestandan enligt de flesta måtten sjunker med 50%, men i vissa fall med så mycket som 90%. Detta tyder på att andra, långsammare metoder inte kan användas i realtid, utan att mer forskning, specifikt inriktad på spårning i realtid, behövs.

In this thesis, findings in three areas of computer vision estimation are presented. First, an improvement to the Kanade-Lucas-Tomasi (KLT) feature tracking algorithm is presented in which gyroscope data is incorporated to compensate for camera rotation. This improved algorithm is then compared with the original algorithm and shown to be more effective at tracking features in the presence of large rotational motion. Next, a deep neural network approach to depth estimation is presented. Equations are derived relating camera and feature motion to depth. The information necessary for depth estimation is given as inputs to a deep neural network, which is trained to predict depth across an entire scene. This deep neural network approach is shown to be effective at predicting the general structure of a scene. Finally, a method of passively estimating the position and velocity of constant velocity targets using only bearing and time-to-collision measurements is presented. This method is paired with a path planner to avoid tracked targets. Results are given to show the effectiveness of the method at avoiding collision while maneuvering as little as possible.