CAW02 - #Hashtag-based donations for (kite)surfers & divers

Savelocal is a South African / Dutch non-profit organisation on a mission to make nature conservation simple and actionable. The way it works is truly innovative: by tagging your posts on Instagram with #savelocal, the company will donate from your bank account a fixed amount to the conservation project that is closest to where the pictures were taken.

Savelocal normally uses the geolocation linked to the post, often though this is not available, and your computer vision skills can come to the rescue;)

The proposed challenge is based on computer vision, optimisation and even API handling (for bonus points) …

 

 

 

When users decide to donate through #savelocal, the company uses Instagram’s geo-location to determine where the photo was taken. However, sometimes, users will forget to pin their location and we are left with an issue: where should the donation go to? Hence, the objective of Savelocal’s Hackathon challenge is to use the image data alongside the tabular data to define two main things.

Firstly, the core purpose of the challenge is image recognition: by using the image data set and labels, the goal is to create a computer vision script that will manage to identify in which locations pictures were taken. Additionally, by using the joined jupyter notebook with scraping code (commented and attached with a README.md file to easily be used during the challenge), participants can also increase the size of the dataset or completely modify it to their will (be as creative as you want).
The second part of the challenge is to use the NGOs file (with their description and location) alongside the image to determine to which NGO the donation should be given. Actual proximity is the most important, but you can make this more advanced by understanding the scene as well. If the picture is of the ocean, focus on marine conservation projects. If there is a specific animal photographed (such as penguins in Boulders Beach), then the donation could go towards animal conservation instead!

Finally, the last (and bonus) task (for those who are brave enough to attempt it), would be to create a live flow by using the Instagram (or Facebook/Twitter) API. Whenever someone posts an image with #savelocal, you fetch the image, compute the closest conservation project and comment its name to the post.

Computer vision expert
Back-end developer
Data scientist
API expert
Mobile developer