The application uses Google Vision API (OCR) to extract text from an image, finds the question, and uses a mixture of Google Web Search API, Bing, and DuckDuckGo to estimate the correct answer.

Questions are answered within 5 seconds depending on internet speed.

Works by counting the number of occurrences of each possible answer from each search engine. If the question is in the form ".. is NOT the .." then it will check the lowest amount of occurrences.

The app displays confidence ratings for each answer and then a best guess.

Works on pub quiz machines, trivia websites and apps.

Doesn't get everything right but from testing gets about 70-80% correct.

Share this project: