Show HN: Challpix – challenge-based, photo sharing app that wants to help AI

This started as a side-project in my last year of university but continued working until present day when it is almost complete. Challpix is a challenge-based, photo-sharing social network whose purpose is to bring back the naive, unfiltered and innocent way of visual expression. (PRish description) When I was building the app, my Image Processing professor said that his department lacked data to train the neural nets and I thought that this is an opportunity to be useful in all those 4 years. We are trying to build a huge dataset of photos that can be mined or used to train neural networks. We found out that the name of challenges can act as noisy labels, each challenge describing a context, not just a keyword like a hashtag (ex. Instagram #s). Nowadays, neural networks are all the fuss and probably you already know that data can improve object recognition, context extraction and many other things. Challpix is trying to share the data with everybody. Academia, NGOs and other not-for-profit entities will be able to access our data free of charge. We are also enterprise-oriented, we planned to offer in the future pay-per-use plans for businesses. Imagine this, you are a NGO that is trying to find all the uncollected garbage around Bucharest and map the location of it. All you have to do is to make a challenge and hopefully some angry people will take photos of the garbage, so the geolocation will be stored. With the data you can build an interactive map that later can be used to make a(nother) point on why the mayor sucks. It’s great, as Trump would have said. Through Challpix you are improving your already smart city (like the Skynet-automated Bucharest) without moving a finger, unbelievable! Think of other examples, like mapping the nests of birds. The app is currently available only for Android: https://play.google.com/store/apps/details?id=com.challpix.luci.android.challpix&hl=en https://play.google.com/store/apps/details?id=com.challpix.l... Every feedback counts. Thank you guys.