Repository of Faculty of Science
Not a member yet
1747 research outputs found
Sort by
Comparison of models for object detection and pose estimation for evaluation of cheerdance jumps
No full textCilj diplomskog rada je usporedba modela detekcije objekata i procjene poza za svrhu pouzdane evaluacije cheerdance skokova. Koristeći tehnike transfer learninga, fino podešavamo YOLOv7, Faster R-CNN, OpenPose i YOLO Pose. Ovo istraživanje pruža ključne uvide trenerima, sportašima i istraživačima, usmjeravajući ih kod odabira modela za evaluaciju skokova. Štoviše, naglašava potencijal korištenja transfer learninga za učinkovitu primjenu modela dubinskog učenja, unaprjeđujući evaluaciju izvedbe cheerdance skokova.This thesis conducts an assessment of object detection and pose estimation models for the evaluation of cheerdance jumps. Employing transfer learning techniques, we fine-tune YOLOv7, Faster R-CNN, OpenPose, and YOLO Pose. This research provides insights for coaches, athletes, and researchers, guiding model selection for jump assessment. Moreover, it underscores the potential of transfer learning to adapt deep learning models effectively, advancing cheerdance performance evaluation
Displays
No full textZaslon je uređaj koji omogućuje vizualni prikaz informacija. U ovom završnom radu objašnjeno je kako su se karakteristike zaslona mijenjale ovisno o pojedinoj vrsti zaslona. Objašnjen je kratki princip rada, odnosno stvaranja slike na CRT zaslonu, plazma zaslonu, LCD zaslonu, LED zaslonu te OLED zaslonu. Nakon čega je u zadnjem dijelu rada opisana tehnologija rada zaslona osjetljivih na dodir.A screen is a device that enables the visual display of information. In this final paper, it was explained how the characteristics of the screen changed depending on the screen. A brief principle of operation, i.e. creating an image on a CRT screen, plasma screen, LCD screen, LED screen and OLED screen, is explained. After that, in the last part of the work, the technology of touch screen operation is described
Surfactant-free microemulsion composed of water, ethanol and heptanol and its effect on silicon dioxide synthesis
No full textMikroemulzije su prozirni, termodinamički stabilni i optički izotropni sustavi koji se sastoje od različitih omjera ulja, vode, surfaktanta i kosurfaktanta. Dugo se smatralo da moraju sadržavati značajnu količinu površinski aktivne tvari. Ipak, dokazano je da se mogu formirati i iz ternarnih smjesa ulja, vode i „amfi-otapala“. Prema faznom dijagramu trokomponentnog sustava voda/etanol/heptanol mikroemulzijskom tehnikom sintetizirao sam nanočestice silicijeva dioksida i njihova svojstva (oblik i veličinu) analizirao pretražnom elektronskom mikroskopijom. Ovim radom želim osvijestiti važnost nanotehnologije u naznovrsnim područjima ljudske djelatnosti i zainteresirati istraživače za daljnji rad na ovom sustavu.Microemulsions are transparent, thermodynamically stable and optically isotropic systems consisting of different ratios of oil, water, surfactant and cosurfactant. For a long time, it was considered that they must contain a significant amount of surfactant. However, it has been proven that they can also be formed from ternary mixtures of oil, water and an "amphi-solvent". According to the phase diagram of the three-component system water/ethanol/heptanol, I synthesized solid silica nanoparticles using the microemulsion technique and analyzed their properties (shape and size) by scanning electron microscopy. With this work, I want to raise awareness of the importance of nanotechnology in various areas of human activity and interest researchers in further work on this system
PLAN UPRAVLJANJA ISTRAŽIVAČKIM PODACIMA
No full textPlan upravljanja istraživačkim podatcima pri projektu Fragilnost fotosinteze u moru uslijed klimatskih promjena (HRZZ IP-2022-10-8859
Hormones and hormonal disorders
No full textHormonska ravnoteža važna je za opće zdravlje čovjeka. Hormoni imaju nezamjenjivu ulogu kada je u pitanju regulacija mnogih fizioloških procesa. Hormonski poremećaji dovode do mnogih mentalnih, fizičkih i emocionalnih problema koji imaju dugotrajan učinak na život čovjeka. Fizički učinci mogu uključivati promjene u metabolizmu, težini, menstrualne nepravilnosti, smanjenu plodnost i abnormalnosti rasta. Emocionalni i mentalni učinci mogu se manifestirati kao promjene raspoloženja, anksioznost, depresija i kognitivna oštećenja. Dugotrajna hormonska neravnoteža može doprinijeti razvoju kroničnih bolesti poput dijabetesa, kardiovaskularnih poremećaja i određenih vrsta raka. Razumijevanje zdravstvenih problema koje sa sobom nosi bilo koji hormonski poremećaj opisan u ovom radu može pomoći u promicanju optimalne endokrine funkcije.Hormonal balance is important for overall human health. Hormones play an indispensable role in regulating many physiological processes. Hormonal imbalances lead to many mental, physical, and emotional problems that have a long-term impact on a person's life. Physical effects can include changes in metabolism, weight, menstrual irregularities, reduced fertility, and growth abnormalities. Emotional and mental effects can manifest as mood swings, anxiety, depression, and cognitive impairment. Long-term hormonal imbalances can contribute to the development of chronic diseases such as diabetes, cardiovascular disorders, and certain types of cancer. Understanding the health problems associated with any of the hormonal disorders described in this article can help promote optimal endocrine function
Automatic assessment of lists in the introductory programming domain
No full textCilj rada je istraživanje mogućnosti automatskog vrednovanja koncepta listi kao temeljnih struktura podataka. Kratki zadaci u području uvodnog programiranja u Pythonu osmišljeni su koristeći Virtual Programming Lab (VPL) alat u Moodle sustavu.The aim of this thesis was to explore the possibilities of automatic assessment of lists as a basic data structure. Short tasks in the domain of introductory Python programming were designed using Virtual Programming Lab (VPL) tool within the Moodle system
Security implications of cookie usage in web applications
No full textU ovom radu definiran je pojam kolačića te su objašnjene njihove osnovne vrste i posebni tipovi poput superkolačića i zombie kolačića. Istraženi su sigurnosni rizici povezani s njihovom upotrebom, kao i preporučene metode zaštite korisničkih podataka. Opisane su različite strategije implementacije kolačića te njihov utjecaj na sigurnost web aplikacija. U praktičnom dijelu rada, kolačići su implementirani u MERN aplikaciju za rezervacije, čime je prikazana njihova uloga u upravljanju sesijama i zaštiti korisničkih informacija.This paper defines the concept of cookies and explains their main types, including special forms such as supercookies and zombie cookies. It explores the security risks associated with their use and outlines recommended practices for protecting user data. Various cookie implementation strategies are described, along with their impact on web application security. In the practical part of the paper, cookies were implemented in a MERN-based booking application to demonstrate their role in session management and safeguarding user information
Gall wasp (Cynipidae) biodiversity on oak of Virgil (Quercus virgilliana (Ten.) Ten)
No full textOse šiškarice (Cynipidae) predstavljaju vrlo raznoliku porodicu unutar reda opnokrilaca poznatu po induciranju morfološki specifičnih biljnih struktura – šišaka. Šiške nastaju kao rezultat složene interakcije između biljke domaćina i kukca pri čemu biljka reagira na prisutnost jajašca ili aktivnost ličinke stvaranjem tkiva koje omogućuje razvoj i prehranu ličinke. Ovaj rad donosi dokumentirano opažanje šišaka izazvanih osama šiškarica na hrastu drmunu (Quercus virgiliana (Ten.) Ten) u području središnje Dalmacije. Ukupno je prepoznato 17 vrsta šiškara, od kojih prevladava rod Andricus, popraćen rodovima Neuroterus, Cynips i Biorhiza. Zabilježeno je 18 oblika šišaka pri čemu su najčešće šiške na pupovima, zatim na listovima, kupulama i cvatovima.Gall wasps (Cynipidae) represent a highly diverse family within the order Hymenoptera, known for inducing morphologically specific plant structures - galls. Galls arise because of a complex interaction between the host plant and the insect, whereby the plant responds to egg presence or larval activity by forming specialized tissue that facilitates the larva’s development and nourishment. This study presents a documented observation of galls induced by gall wasps on the Italian oak (Quercus virgiliana (Ten. Ten) in the region of central Dalmatia. A total of 17 gall wasp species were identified, with the genus Andricus being the most dominant, followed by the genera Neuroterus, Cynips and Biorhiza. Eighteen distinct gall morphotypes were recorded, with the most frequent occurring on buds, followed by those on leaves, cupules and inflorescences
Analysis and application of SEO and GEO methods to improve the visibility of web applications
No full textOvaj završni rad bavi se obradom temeljnih principa optimizacije za tražilice (SEO), kao i novije paradigme generativne optimizacije (GEO), koja se odnosi na vidljivost web stranica u rezultatima pretraživanja temeljenima na umjetnoj inteligenciji. Teorijski dio rada uključuje pregled tehničkih, sadržajnih i lokalnih SEO metoda, kao i ulogu semantičkog HTML-a, pristupačnosti i strukturiranih podataka. U praktičnom dijelu izrađena je jednostavna web aplikacija u React okruženju, na kojoj su prikazane konkretne implementacije SEO i GEO strategija. Evaluacija učinka provedena je pomoću alata kao što su Rankscale.ai i Backlinko SEO tester, uz usporedne metrike prije i nakon optimizacije. Rad završava analizom postignutih rezultata i osvrtom na ograničenja i mogućnosti daljnjeg razvoja.This final thesis explores the fundamental principles of Search Engine Optimization (SEO) and introduces the concept of Generative Engine Optimization (GEO), which focuses on the visibility of websites within AI-driven search results. The theoretical part includes an overview of technical, content-based, and local SEO methods, as well as the role of semantic HTML, accessibility, and structured data. In the practical segment, a simple React-based web application was developed to demonstrate the implementation of selected SEO and GEO strategies. The effectiveness of the applied techniques was evaluated using tools such as Rankscale.ai and Backlinko SEO Tester, with a comparative analysis of performance metrics before and after optimization. The thesis concludes with a discussion of results, limitations of the approach, and suggestions for future work
Stress detection using galvanic skin response signal
No full textZnačajan utjecaj stresa na mentalno i fizičko zdravlje čini detekciju stresa važnim istraživačkim područjem. Poznavanje načina na koji pojedinac reagira na stres može pomoći u smanjenju rizika od izgaranja, povećanju učinkovitosti i općem unapređenju kvalitete života. Identifikacija fizioloških pokazatelja pobuđenosti organizma stoga bi bila posebno korisna u radnom i obrazovnom okruženju. Ovaj rad istražuje upotrebu signala galvanskog odziva kože (galvanic skin response, GSR) kao metode za detekciju različitih razina pobuđenosti. U tu svrhu osmišljen je eksperiment koji uključuje različite zadatke namijenjene izazivanju stresa. U istraživanju je sudjelovao 41 ispitanik. Iz svakog mjerenog signala izdvojeno je 13 značajki, a uz to je zabilježena i subjektivna procjena doživljenog stresa. Provedena je statistička analiza kako bi se identificirali značajni obrasci u podacima te istražile moguće razlike u reakcijama na različite razine stresa. Kako bi se procijenila uspješnost razlikovanja različitih razina pobuđenosti na temelju GSR signala, korišteni su Support Vector Machines (SVM) i Linear Discriminant Analysis (LDA) modeli strojnog učenja. Rezultati pokazuju da GSR uspješno odražava prisutnost pobuđenosti. Međutim, na temelju prikupljenih podataka nije moguće razlikovati sve razine stresa.Stress detection is an important research area due to the significant impact of stress on mental and physical health. Understanding how an individual responds to stress can help reduce the risk of burnout, improve efficiency and support overall well-being. The identification of physiological arousal would, therefore, be particularly useful in workplace and educational settings. This work explores the use of the galvanic skin response signal (GSR) as a method for detecting varying levels of arousal. For this purpose, experiment involving different stressinducing tasks was designed. Forty-one participants took part in the study. From each recorded signal, 13 features were extracted and subjective measure of experienced stress was determined. Statistical analysis was done to identify significant patterns in the data and uncover potential differences in responses across varying levels of stress. At the end, Support Vector Machines (SVM) and Linear Discriminant Analysis (LDA) machine learning models were utilized to assess the practical potential of distinguishing stress levels based on GSR signal variations. The results show that GSR successfully reflects the presence of arousal. However, it was not possible to distinguish all stress levels from the data