During the past week, I mainly worked on adjusting logic for interpreting CV output as we have moved onto a new duo camera set up as our main counting mechanism. The old logic for interpreting output was very erroneous due to different camera angles, and it took me a decent amount of time to figure out why. After running the two camera setup on my own computer, I found that the video feed is laggy as my computer does not have enough processing power to process the two video feed, therefore we need to find other solutions to perform the processing in realtime. I previously migrated some processing and CV code to google colab in order to process test video footages faster, and it was able to perform better than my local computer with the GPU environment on google colab, so after discussion with our team, we have decided to run the backend processing on google colab via streaming. During the week, David and I did some testing and we were able to stream frames from raspberry pi to google colab through sending the frame to a webserver, and fetching the frame based on a get request on google colab. The next steps will be integrating it into our existing backend code, which I plan to finish by tomorrow. We will test our system as a whole on Sunday to make sure that our main CV backend works properly. Next week will be dedicated to reducing errors in our backend and try to run validation tests.