Team name / Company name: Student Dreamers
Team leader: David Zahound
Challenge: no. 5: Longer life for furniture
Problem: IKEA is growing. We can observe growth in customer orders over 12 % each year. This also means that the number of orders where a part is missing is growing too. Most orders are fine, but sometimes a screw goes missing, some little part cracks or, let’s be honest, we break a part. Thankfully, IKEA sells replacements, but the customer experience when ordering a new part is overwhelming. This can lead to increased load of requests from unhappy and stressed customers at the support.
Solution: We simplified the process of ordering such parts. Our mobile-friendly web app gives users an option to search for replacement parts for their products using both text search and, because, let’s be serious, you don’t remember the name of your nightstand from IKEA and the box is long gone, an innovative AI detection of their product using their phone’s camera.
Impact: There are almost 450 IKEA stores on this planet. Roughly, they sell almost 10 thousands products to over 2 billion people. If we can save 2 % of furniture bought each year we can save over 100 square kilometers of forests and lower the cost of customer support required.
Feasibility: IKEA can kill two birds with one stone. In the first place customers will find a much easier way to acquire missing parts. And a second, the app can relieve a lot of cost-heavy customer service from call centers and stores.
What you built: We trained a neural network based AI with the AED20K dataset and our own photos of IKEA furniture. Then we built a web app with focus on mobile phones first, implemented the AI and added support for streamed data (live video stream from a camera), gathered a list of products from IKEA’s website and linked it all together. https://github.com/student-dreamers/greenhack-camera-react https://github.com/student-dreamers/greenhack-camera
What you had before: We started working on Friday afternoon and we’ve created an app totally from scratch. Only thing we’ve used was a pre-trained furniture detection model which we have enhanced.
What comes next: We want to fully integrate the app with IKEA’s catalog and improve the AI detection using 3D furniture models which IKEA is already using. There shouldn’t be any notable obstacles since we’re not building a downloadable app, users can run it from their phone’s web browser, and we do not affect products in any way.