ReSoul

Inspiration

ReSoul was created with three central pillars in mind: sustainability, wellness, and business integration. Repairing climbing shoes means less wasted equipment in landfill. ReSoul enables novice and experienced climbers alike to invest in their wellness through rock climbing. As an excellent source of exercise that can easily be done inside or outdoors, rock climbing has become a crucial way for many to relax, play, and connect with others. By providing a digital platform for users to identify local repair services, we are providing a more seamless retail experience that reaches a broader customer base.

What it does

Over time, the rubber used to give climbing shoes their grip naturally wears down. Sadly, your climbing shoes won't last forever, but with ReSoul, you can make them last a little longer. Using ReSoul, you can upload a photo of your climbing shoes and our AI-trained image recognition model will identify if your shoes require resoling. Resoling is the process of replacing the worn rubber on the front and/or bottom of your climbing shoes. Resoling can prolong the lifetime of your favourite climbing shoes by months or even years, and is a worthwhile investment in your quality climbing apparel. Before & after resoling climbing shoes

How we built it

Using convolutional neural network (CNN), we trained our machine learning model with hundreds of climbing shoes pictures that needed repair or no repair needed. With evaluation and testing, this model was trained to a satisfactory accuracy to be used for our climbing shoes analyzer. We used Flask to integrate back-end with React front-end.

Challenges we ran into

Some pictures we used for the machine was corrupted and damaged our results findings. It was difficult to finally spot out 3 pictures that were the cause of this.
As all of our first time working with Flask, it was difficult to find a secure connection to receive uploaded images into a folder that our machine learning model can access.

Accomplishments that we're proud of

We created a functional convolutional neural network image classification system with a 56% accuracy rate
It was some of our first time using React and Figma, and we managed to create a beautiful website and an even more beautiful mock-up!

What we learned

We should have used containerization in our machine learning model to allow the code to run easily on different computers within the same environment
We learned the basics of React and Figma