Pregled bibliografske jedinice broj: 1265721
System for Recognition and Evaluation of Handwritten Arithmetic Expressions
System for Recognition and Evaluation of Handwritten Arithmetic Expressions // 8th Croatian Computer Vision Workshop (CCVW 2019)
Zagreb: Sveučilište u Zagrebu, 2019. str. 1-6 doi:10.20532/ccvw.2019.0008 (poster, recenziran, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 1265721 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
System for Recognition and Evaluation of
Handwritten Arithmetic Expressions
Autori
Galić, Vlado ; Hrkać, Tomislav
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Skup
8th Croatian Computer Vision Workshop (CCVW 2019)
Mjesto i datum
Zagreb, Hrvatska, 23.10.2019
Vrsta sudjelovanja
Poster
Vrsta recenzije
Recenziran
Ključne riječi
mathematical expression recognition, deep learning
Sažetak
Recognition of mathematical expressions today is a very interesting area of application of deep learning. The problem is quite complex and requires a complex system to solve it. The problem becomes more complex as arithmetic expressions are handwritten. The use of convolutional neural networks yields satisfactory results but leaves some room for improvement. In this paper, a simplified variant of the above problem is solved by using an approach that consists of extraction and classification of individual symbols using convolutional neural networks followed by syntactic parsing of the expression taking into account the symbol positions.
Izvorni jezik
Engleski
Znanstvena područja
Računarstvo
POVEZANOST RADA
Ustanove:
Fakultet elektrotehnike i računarstva, Zagreb
Profili:
Tomislav Hrkać
(autor)
