An mobile-based application that allows users to take a picture of certain event flyer and help user save corresponding date, time and location for the event in accordance to information provided in the taken photograph.
Our inspiration came from our daily experience that when we are trying to record the events we saw from the posters, we need to type them manually into Google Calendar. And that’s a lot of work! We need to choose the date, choose the location, and choose the time. Imagine when you see ten interesting events’ poster on the same wall, that’s gonna take you about 15 minutes to record all the events. In order to save more time for busy^100000 college students, we decided to develop SnapCal, an app which helps you add events to Google Calendar with a simple snap!
With a simple snap, you can have the poster event saved to your calendar. As you take a picture of the event poster, SnapCal automatically detects the texts, identify the event title, dates and times. The correct event and location will be added to your Google calendar. You don’t need to input anything, and SnapCal will handle everything for you.
We utilized the Google Cloud Vision API to extract texts from images. After that, we identify the title based on the position and size of texts. The date and time are matched out using patterns. The texts are also compared with a building list, and then sorted by the Google Maps API.
Filtering the date and time is a challenging task, since there are so many formats.
For now the application successfully detects location on campus at an accuracy of approximately 80%. For each picture of poster that we test against the application, date and time for corresponding events despite the fact that texts and images vary in form, shape and color for each poster test. We also successfully create corresponding event in user’s google calendar with efficiency. The operations required is easily-understood and within click.
- Applying Google API, IBM API, etc.
- Basic text matching algorithm
- Data extraction from free-form text
- UI interface design knowledge
- Improve information detecting accuracy especially location matching to enlarge the scope of usage for the app from Cornell Campus to any valid location.
- Implement actual app interface for SnapCal in order to provide valid using channel for user