Team Status Report 10/26

Summary

We made progress over the last week on multiple fronts:

• The server was dockerized and a production and docker variant of the container was created.
• A basic version of authentication was formally decided on as an MVP.
• The YOLO model was tested via new validation scripts and was ensured to meet MVP standards for the presentation. Further research is being done to integrate the Grounded Dino openset model.
• CAD models were created and printed with 3D printers.
• The server and RDS database integration had significant work completed and the kinks should be finished off next week.

What are the most significant risks that could jeopardize the success of the project? How are these risks being managed? What contingency plans are ready?

The biggest risk is that our ML model somehow doesn’t generalize to the presentation room while doing fine in other environments. It’s being managed by getting a wide array of testing environments in the training dataset so it can generalize to any reasonably well lit environment.

Although we have a basic MVP with most things integrated, there is still the need to prove out scaling. We will likely be looking on scaling on our own computers via multiple threads etc as an MVP and then potentially transition to the cloud if we have time. Also note that it seems that Rasp Pis might be able to run a slower version of our model locally so perhaps cloud wont be needed at all.

We have still yet to make significant progress on the audio component although it is comprehensively planned out. There are many parts to the equation and although we predict each one to have a high chance of success the only way to figure it out is to test. We will likely start creating an MVP this week and finish by next week.

Were any changes made to the existing design of the system (requirements, block diagram, system spec, etc)? Why was this change necessary, what costs does the change incur, and how will these costs be mitigated going forward?

There have been no major changes to the design of the system other than add ons like adding Grounded Dino to make the already somewhat robust YOLO model even more accurate.

The change does incur computation costs though MVP requirements will still be met if Grounded Dino is not integrated.