Meeting the accuracy requirements for tracking and prediction remains the foremost risk to the success of our project. At the outset of our project, we recognized that acquiring sufficient information to train our predictive model within the semester’s timeframe could be challenging. Consequently, we decided to augment our occupancy data with Hammerschlag class schedules and synthetic samples. Our primary objective remains to refine the construction of our predictive component to achieve our 80% estimation target gradually.
Moreover, in our testing this week, we also encountered some difficulties with our computer vision pipeline in capturing individuals walking in close proximity to the camera. The object detection mechanism exhibited inconsistency while monitoring the nearest doorway to the camera, resulting in inaccurate estimations of room occupancy. To rectify this issue, we plan to conduct further testing of our system in the upcoming week, with adjustments to our camera placement and detection boundaries, to pinpoint the source of the problem in our system.
During this week, we have not made any changes to the existing design of the system or our schedule. We remain mostly on schedule with the subsystems of our product, and will be preparing to demo a limited version detection and predictive systems for the interim demo next week.