Inspiration

Workshop #4: Climeworks & Accenture

PrimeClime

The Climate Crisis – we often speak about it – but also we often don't know what is the right choice to make in order to live in an eco-friendly way. Some people really care a lot, but again others have too few incentives to change their habits. So we thought it would be great to seek for a solution, where everyone would enjoy acting right for our planet. And what is better than a game to encourage everyone to join in and win?

We thought that using gamification and "challengification", we could develop an app that gives you "green reward points" for good eco-deeds. With an approach similar to the Bike To Work Challenge, we wanted to give you the opportunity to participate in different challenges. Weekly challenges, challenges with your friends or family, public ones and many more would create a big incentive to participate, mostly because we are all competitive in a certain way.

With this background idea, we also thought that using in-app purchases, ads and other tools we could raise money for Climework's mission. Selling those "green reward points" would directly help Climework to bind CO2 from the atmosphere. And simultaneously, we give the people a better feeling, they see and feel that they acted right and had an impact on the environment. It is important, that people start to realize that their decisions do have an impact on a global scale, so why not show them? With PrimeClime, this could be possible!

How we built it

The idea is the following: Using an app with user account, we let the user input his deeds and acts. For the beginning, categories like consumption, transportation and active CO2 binding are the main target. Using CO2 footprint databases, we can calculate how much CO2 is produced by a user based on his inputs. To prevent users to input only very eco-friendly behaviour, we created a system where all inputs result in reward points, but the less CO2 is produced by the input, the more reward points are earned.

For the prototype, we focused on alimentary products of Coop: Using an image of a Coop-receipt, we extracted text data specifying product type and quantity in order to calculate the purchased product CO2 emissions. We compared the products with a database of approximately 200 basic alimentary products and their CO2 emissions per kg, all that hosted on a server. Using the quantity, we calculated each single product's CO2 emissions and assigned them to a reward point score, which was then uploaded on the user account.

Challenges we ran into

Challenges were not rare, but given the short time we had we were able to master some of them:

Starting at the beginning, extracting text data of a receipt – which is often very small printed and faded – was not easy. Even after cleaning up using some filters, product names and quantity were sometimes not correctly extracted. Also, there were problems with the non-unitary format of the product names and quantity. Some products are given by simple name and weight in kg one one line, others have special names and quantities like "1" or even sometimes, quantity or weight is given in the product name. Surely, with more time and resources, these problems can be solved. Using the bar code and the encoded receipt information would be a solution, but for that, we would have to work closely with the retail company, as we would need access to their product databases. Then, we had some difficulties to compare our basic alimentary database entries with the text data we had extracted. Sometimes, a false character in the product name would defeat our comparison algorithm. Also, to handle all the small irregularities in special names or compound names was not trivial. But at the end, we had a good approximation of the CO2 emissions caused by the purchased aliments.

Accomplishments that we're proud of

Participating for the first time at a Hackathon, we were very proud of what we accomplished overall! Already extracting, treating and cleaning up the raw text data from a receipt was a big accomplishment. Also handling different image types like .HEIC/HEIF, .JPEG etc was nice.Then the comparison and rewarding system uploading a point score to the user account was a milestone too. Also, we are very proud of our idea, which could – with some more time and effort – become an entertaining solution to help save our world.

What we learned

We have definitely learned a lot! The text extraction was new to us, we learned a lot about filtering and bug fixing! In general the Hackathon was something great to experience!

What's next for PrimeClime

PrimeClime is a very interesting project, but as we are all mid studies, we would like to improve and further develop the prototype but surely not as intensively as the last few days!

Share this project:

Updates