| Abstract [eng] |
These master theses consist of three main logical sections, which include the analysis of analytical problem technologies and design methodology; the specification of project requirements and project implementation; the research of system quality; the implementation of suggested improvements and finally the design and realization of experiment. The first part includes the study of document classification problem, relative technologies and optical symbol recognition systems. In the analysis phase the project itself is introduced. It is the system that presents the information about additives found in food from a given food label photo. Initially, it is important to gather up the requirements and relative technologies for the project implementation. The second part includes the specification of requirements for the project and project implementation. It was decided to use Microsoft .NET technology and client – server architecture. The third part includes the tests of system quality. The most important criterion of the system quality is correctness, which shows how accurately the system finds the additives in food label. Using testing data system gives only 49% of correctness, which means it finds only half of all product label additives. For this reason there were improvements suggested and later on they were implemented. During the experiment with enhanced system version the percent of correctness grew up to the 88%, which completely justify the expectations. |
