prikaz prve stranice dokumenta Čitanje s usana primjenom računalnog vida
No public access
master's thesis
Čitanje s usana primjenom računalnog vida
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. urn:nbn:hr:168:968375
Arambašić, Lucija
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Arambašić, L. (2022). Čitanje s usana primjenom računalnog vida (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:968375

Arambašić, Lucija. "Čitanje s usana primjenom računalnog vida." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:968375

Arambašić, Lucija. "Čitanje s usana primjenom računalnog vida." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2022. https://urn.nsk.hr/urn:nbn:hr:168:968375

Arambašić, L. (2022). 'Čitanje s usana primjenom računalnog vida', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 22 November 2024, https://urn.nsk.hr/urn:nbn:hr:168:968375

Arambašić L. Čitanje s usana primjenom računalnog vida [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2022 [cited 2024 November 22] Available at: https://urn.nsk.hr/urn:nbn:hr:168:968375

L. Arambašić, "Čitanje s usana primjenom računalnog vida", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2022. Available at: https://urn.nsk.hr/urn:nbn:hr:168:968375

Metadata
TitleČitanje s usana primjenom računalnog vida
Title (english)Lip Reading Using Computer Vision
AuthorLucija Arambašić
MentorZoran Kalafatić (mentor)
Committee memberZoran Kalafatić (predsjednik povjerenstva)
Committee memberVladimir Čeperić (član povjerenstva)
Committee memberTomislav Hrkać (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2022-07-15, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Ovaj rad bavi se rješavanjem problema čitanja s usana. Problem je formuliran kao višeklasna klasifikacija slijeda, gdje se video snimka označava slovom koje je na video snimci izgovoreno. Na snimkama se izgovaraju slova engleske abecede. Napravljena je analiza koja slova je nemoguće razaznati jedna od drugih te je problem reformuliran tako da su ta slova grupirana te označena istom oznakom. Potom se koristio predtrenirani SqueezeNet model za ekstrakciju značajki iz video sličica. Ekstrahirani sljedovi značajki su potom klasificirani uporabom tri različita modela: povratnom neuronskom mrežom, logističkom regresijom i strojem potpornih vektora.
Abstract (english)
This thesis tackles the problem of automated lip-reading. The problem is formulated as a multiclass sequence classification task, where each video is considered a sequence of images and is labeled with the letter that is uttered in the video. The letters spoken in the videos are English alphabet letters. An analysis of letter utterances that cannot be distinguished between each other using only visual information is conducted. The problem is then reformulated by grouping those letters and labeling them with one joint label. Afterward, a pre-trained SqueezeNet model was used for image feature extraction and the resulting feature sequences were classified by three different models: recurrent neural network, logistic regression, and support vector machine.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:168:968375
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2023-01-14 08:19:50