Sirvi Autor "Anbarjafari, Gholamreza, supervisor" järgi

Nüüd näidatakse 1 - 7 7

listelement.badge.access-status Avatud juurdepääs ,
Augmented Reality Card Game Based On ArUco Marker Detection
(Tartu Ülikool, 2019) Lepik, Tõnis; Haamer, Rain Eric, supervisor; Anbarjafari, Gholamreza, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
ArUco is a fiducial marker detection library that uses a square marker system for identifying different patterns with unique values. This thesis explores the possibility to use those markers as an AR element in video games so, that any physical marker may represent any virtual object that is assigned to it in the software. Such system could be used for purposes, where the cost or volume of game specific cards is too high. First and second part of the thesis give a brief overview of AR/VR-, mobile- and overall video gaming industry. Third part looks into fiducial marker detection and more specifically ArUco technology. Fourth part describes the development process of a mobile application and fifth part presents the results of the carried out user testing.
listelement.badge.access-status Avatud juurdepääs ,
Automatic Speech-based Emotion Recognition
(Tartu Ülikool, 2018) Hook, Joosep; Anbarjafari, Gholamreza, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
The main objectives of affective computing is the study and creation of computer systems which can detect human affects. For speech-based emotion recognition, universal features offering the best performance for all languages have not yet been found. In this thesis, a speech-based emotion recognition system using a novel set of features is created. Support vector machines are used as classifiers in the offline system on Surrey Audio-Visual Expressed Emotion database, Berlin Database of Emotional Speech, Polish Emotional Speech database and Serbian emotional speech database. Average emotion recognition rates of 80.21%, 88.6%, 75.42% and 93.41% are achieved, respectively, with a total number of 87 features. The online system, which uses Random Forests as it’s classifier, consists of two models trained on reduced versions of the first and second database, with the first model trained on only male samples and the second trained on both. The main purpose of the online system was to test the features’ usability in real-life scenarios and to explore the effects of gender in speech-based emotion recognition. To test the online system, two female and two male non-native English speakers recorded emotionally spoken sentences and used these as inputs to the trained model. Averaging over all emotions and speakers per model, it is seen that the features offer better performance than random guessing, achieving 28% emotion recognition in both models. The average recognition rate for female speakers was 19% in the first and 29% in the second model. For male speakers, the rates were 36% and 28%, respectively. These results show how having more samples for training for a particular gender affects emotion recognition rates in a trained model.
listelement.badge.access-status Avatud juurdepääs ,
Deep Learning Based Automated Job Candidate Interview Screening
(Tartu Ülikool, 2019) Aktas, Kadir; Anbarjafari, Gholamreza, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
Traditional way of recruitment process is challenging for both the candidate and the employer. To apply for a job, the candidate needs to prepare a CV. On the other hand, the employer needs to check all the submitted CVs and analyze the candidate data manually. These aspects can make the process very time consuming, especially when there are many candidates. Furthermore, the manual analysis of the candidate data is very open to human bias. The thesis proposes an automated video interview analysis system, which eliminates the problems mentioned above.
listelement.badge.access-status Avatud juurdepääs ,
Image-based automated leaf area calculation using OpenCV
(Tartu Ülikool, 2019) Demišin, Kirill; Kollist, Hannes, supervisor; Anbarjafari, Gholamreza, supervisor; Pedor, Jenni Katri, supervisor
In English: Analyzing RGB images with the help of computer vision is an affordable and reliable tool to study plants. This approach is becoming widely used due to the low cost of high-resolution RGB cameras and the existence of open source software. The main purposes are phenotyping, development observation and detection of illnesses among different crop types for both scientific purposes and breeding. Nowadays, leaf area extraction, leaf segmentation and 3D reconstructions of plants are popular problems solved with image processing and machine learning. This thesis describes an algorithm for automated leaf area calculation in captured images. It explains the program’s features and compares the results with the calculations of specialists. Finally, a usable GUI application is designed to help Tartu University lab workers with automation of a manual task. Eesti keeles: RGB kujutiste analüüsimine arvutinägemise abil on taskukohane ja usaldusväärne vahend taimede tuvastamiseks ja nende omaduste hindamiseks. Seda lähenemist kasutatakse laialdaselt kõrge resolutsiooniga RGB-kaamerate madalate kulude ja avatud lähtekoodiga tarkvara olemasolu tõttu. Peamised eesmärgid on taimede fenotüpiseerimine, arengu jälgimine ja haiguste avastamine erinevate kultuuriliikide vahel nii teaduslikul eesmärgil kui ka aretamiseks. Tänapeval on lehtede ala ekstraheerimine, lehtede segmenteerimine ja taimede 3D rekonstrueerimine populaarsed probleemid, mis on lahendatud pilditöötluse ja masinõppega. Käesolevas töös loodi automaatselt toimiv lehtede pindala arvutuse algoritm, mis kasutab pildistatud piltide ja arvutab lehtede pindala. Töö selgitab programmi omadusi ja võrdleb tulemusi spetsialistide arvutustega. Lõpuks on kasutatav GUI rakendus mõeldud selleks, et aidata Tartu Ülikooli taimsete signaalide uurimisrühma töötajatel manuaalülesande automatiseerimist.
listelement.badge.access-status Avatud juurdepääs ,
Imaging Simulator and Geometric Image Stitching for a Low Earth Orbit Satellite with High Spin Rates
(Tartu Ülikool, 2019) Haamer, Rain Eric; Sünter, Indrek, supervisor; Anbarjafari, Gholamreza, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
ESTCube-2 is a low-earth orbit satellite with the main mission of deploying and testing an electrically charged tether and a secondary mission of photographing ground vegetation. The tether deployment and maintaining its separation from the satellite requires a very high spin rate, which poses too many challenges to the imaging system for it to be viable in its current state. In an attempt to resolve this issue, a novel image morphing and stitching algorithm was developed that uses geometric mapping for reconstruction. The proposed method was tested on a specifically made simulation environment that mimics predefined camera parameters. The stitching algorithm was verified to perform within acceptable margins even when expected high spin distortion models were applied.
listelement.badge.access-status Avatud juurdepääs ,
Investigation and Comparison of Kinetostatic Performance Indices for Parallel Mechanisms
(Tartu Ülikool, 2019) Sellis, Ott; Anbarjafari, Gholamreza, supervisor; Daneshmand, Morteza, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
For as long as we have used robots there has also been ongoing research to allow us to use and improve efficiency of automation in our daily lives. As our knowledge about robots has largely improved, so has the complexity of their structures. Thus, various methods and indices have been developed to help designers and engineers determine the best manipulator for a specific task. In addition, the interest towards parallel manipulators has seen growth in the last couple of years due to significantly better performance in various areas in comparison to serial mechanisms. However, no global performance index to evaluate accuracy and allow comparison in that perspective between parallel mechanisms has been developed. This thesis focuses on giving an overview on the developments towards finding a robust kinematic sensitivity index to measure accuracy performance of parallel manipulators.
listelement.badge.access-status Avatud juurdepääs ,
Online Battery Cell State of Charge Estimation for use in Electric Vehicle Battery Management Systems
(Tartu Ülikool, 2018) Dreija, Kristaps; Anbarjafari, Gholamreza, supervisor; Avots, Egils, supervisor; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Tehnoloogiainstituut
The electric vehicle (EV) is a complex, safety-critical system, which must ensure the safety of the operator and the reliability and longevity of the device. The battery management system (BMS) of an EV is an embedded system, whose main responsibility is the protection and safety of the high-voltage battery pack. The BMS must ensure that the requirements for health, status and deliverable power are met by maintaining the battery pack within the defined operational conditions for the defined lifetime of the battery. The state of charge (SOC) of a cell describes the ratio of its current capacity (amount of charge stored) to the nominal capacity as defined by the manufacturer. SOC estimation is a crucial, but not trivial BMS task as it can not be measured directly, but must be estimated from known and measured parameters, such as the cell terminal voltage, current, temperature, and the static and dynamic behavior of the cell in different conditions. Many different SOC estimation methods exist, out of which (currently) the most practical methods for implementing on a real-time embedded system are adaptive methods, which adapt to different internal and external conditions. This thesis proposes the use of the sigma point Kalman filter (SPKF) for non-linear systems as an equivalentcircuit model-based state estimator to be used in one of the current series production EV projects developed by Rimac Automobili. The estimator has been implemented and validated to yield better results than the currently used SOC estimation method, and will be deployed on the BMS of a high-performance hybrid-electric vehicle.