23 Apr 22 – Team Status Report

We made quite a bit of progress this week with our project, and are actually mostly prepared for a demo already 😀 We will focus our efforts on simple incremental changes + lots of testing.

We have thus far:

1. 1. Integrated the entire system with our tablet + Jetson
   2. Completed a benchmark accuracy and speed test
   3. Did extensive full-system bug-testing and fixed some bugs in our back-end code with returns and made the system more robust against incorrect quantity data (e.g. removing more fruits than there are in the fridge)
      
   4. Completed training of self-collected classes, including bell peppers, and are potentially looking at adding more classes.
   5. Debugged and fixed classification speed issues on the Jetson
   6. Spray-painted the background white to improve algorithm accuracy
   7. CV algorithm is now more robust against “false triggers”
   8. Nutritional information now available to user

That being said, we have found out some interesting findings and possible improvements that we can make to our project that will make our product even more refined:

Adding email or in-app notifications to remind the user that their food is expiring soon (backend + frontend)
1. Quantity detection using simple white background thresholding and GrabCut (CV, backend interface already in place)
2. Different fruit type selection (see Figure above) (frontend + CV)
3. Smart recipe detection
4. EVEN MORE TESTING TESTING TESTING.

We are currently comfortably ahead of schedule but will keep pressing on in refining our final product 🙂