A Software Prototype for Assisting the Visually Impaired in Shopping

Uobičajene svakodnevne aktivnosti većini ljudi ne predstavljaju problem, no osobama s određenim poteškoćama mogu značiti prepreku i ograničiti njihovu razinu samostalnosti. Za slabovidne osobe jedna je od tih aktivnosti kupovina. Cilj ovog rada bio je razviti prototip programskog rješenja za pomoć slabovidnim osobama pri kupovini koje se temelji na primjeni umjetne inteligencije i strojnog učenja. Prototip je izveden u četiri faze. U prvoj fazi prikupljeni su i specificirani zahtjevi primjenom metoda introspekcije, proučavanja tehničke dokumentacije, oluje ideja i prototipiranja. Zatim je, u drugoj fazi izrađen prototip visoke vjernosti u alatu za izradu prototipa Figma. Treća faza donijela je izradu funkcijskog prototipa, odnosno implementiranja dijela odabranih funkcionalnosti korištenjem alata za izradu korisničkih sučelja Flutter. U četvrtoj fazi prototip visoke vjernosti evaluiran je testiranjem i ispunjavanjem upitnika o korisničkom iskustvu. Rezultati evaluacije bili su uglavnom pozitivni. Zaključeno je kako je ovo korisno rješenje čiji puni potencijal tek treba biti ostvaren.The usual everyday activities for most people don't represent an issue, but can represent an obstacle for people with certain disability and limit their level of independence. For visually impaired people, one of these activities is shopping. The objective of this thesis was to develop a software prototype for assisting the visually impaired in shopping based on application of artificial intelligence and machine learning. The prototype was developed in four phases. In the first phase the requirements were gathered and specified by applying methods of introspection, technical documentation study, brainstorming and prototyping. After that, in the second phase the high-fidelity prototype was developed in prototyping tool Figma. The third phase brought development of a functional prototype, respectively implementation of chosen functionalities by using Flutter, a tool for creating user interfaces. In the fourth phase the high-fidelity prototype was evaluated by testing and filling out a user experience questionnaire. The results of evaluation were mostly positive. It is concluded that this is a useful solution which full potential is yet to be accomplished