Ready for the last hackathon? We surely hope so. Because this time we have a US $2000 prize that we can’t wait to award! 🙂 That is if you can design a solution that will automatically classify pictures as required below and submit it by next Friday, May 23rd, 24:00.
Before detailing the problem, here’s a short recap for those who are new at this: make sure you read the terms and conditions and follow the steps needed to enter the race. Don’t forget to register your team first using this form.
Classification System for Pictures
Managing pictures is a nightmare. Everyone knows how frustrating it is to locate a particular picture in a very large collection. But this is where you could lend a hand. We need you to dramatically simplify the process by building an automatic classification system. The system should be able to read a collection of pictures and classify them, based on the picture information (Exif and the actual picture content).
- You can write the app in any programming language you like.
- You must accompany your app with some docs. Give us some heads-up, it will support your project.
- Do not forget to mention the name of your team in the docs. That way, we can easily relate to your registration info.
- Check that your Bitbucket repository is private and that you provide us access to view your project.
You must define a set of rules that will include at least the following:
- number of people in the picture
- actual people recognized in the picture (by comparison to known models with face recognition)
- place and the actual conditions (geo location, outdoor/indoor)
- landscape type
- event (by relating to other pictures in the collection)
- time of day
- picture quality
All this information must be written to a database and the database must be easily searchable.
As always, don’t forget that you can use the comments area to ask for info that could help you develop your solution. 🙂
Good luck everyone!