Recognize

When people write down links that they expect you to copy and type out, it's undoubtedly a chore. We decided to take advantage of the excellent OCR algorithms in existence today to fix this problem. Through the power of Microsoft Cognitive Services, we were able to create an app that allows you to search for text fields and links in everyday environments. The app is still in production even as we submit. There’s a bunch we want to add to it. First is sending higher quality images to Microsoft. The way the app is right now, the images being sent aren’t of high enough quality to fully take advantage of Microsoft’s API. We also want to develop better algorithms for determining links. Right now we basically return the user a choice of every single word the API recognized and they get to choose what they want to put onto their clipboard. We want to remove this added hassle and fully automate a link recognizing app.

Along the way we had a lot of hardships. We weren’t Android devs. I remember at one point, we spent around 3 hours trying to make a POST request to Microsoft to use their API. We also routinely ran into problems with asynchronous tasks (especially considering the POST request is asynchronous). Despite all this, we made it and got the app (partially) completed and we couldn’t be more proud.

Built With

• android
• microsoft-cognitive-services
• rest-api