During this week, I made some wood fixtures for mounting the cameras per the professors suggestion so we do not have to tape the cameras onto the walls. The fixtures are convenient as now we can adjust camera positions a little easier, it is also useful for our final demo in May, as we can move part of our setup to our demo location with less effort. As a team, we did more testing on counting accuracy with our two camera setup along with another two cameras for validation. We will try to do more testing in the upcoming days to fine tune our project. From reviewing the testing footages in attempt to find sources of inaccuracy, I found that some errors are due to inherent CV capabilities, such as inaccuracies of bounding boxes during tracking, especially when there are high occlusion. I cannot seem to find good ways to go around that after experimenting with multiple camera angles, etc. However, it seems like our product is having around 80 percent accuracy as for most cases, people are going in and out of rooms one at a time and without much occlusion, and our system is handling those cases well. I am also attempting to refactor parts of the backend code to make it easier to change parts of the logic to adapt to different camera setups.