Parking at the University at Buffalo's North Campus can be extremely limited at times; students, faculty, and visitors often drive around aimlessly and hope to find parking. As a result, students are late for their classes, accidents are more likely to occur, and tensions arise between drivers over parking. Reducing parking lot congestion will alleviate these issues. The goal of our proposed system is real-time tracking of parking lot capacities so that drivers are informed and avoid full lots, thereby reducing congestion. The core of our solution is a machine-learning model that receives video input and detects the number of vehicles in the frame. The system's physical components are a USB camera, a Raspberry Pi 4, and an LCD display. The plan is to mount the system to light poles near parking lots for elevation and better camera coverage. For larger lots, additional cameras can be added. All processing is done locally, which protects the privacy of drivers and their vehicles. In addition, the parking data can be stored and analyzed for trends. Then, Parking and Transportation Services can use this data to improve the parking situation on North Campus further.

Working with model results

Ultralytics

LCD Display

RPLCD Docs

• not creating virtual env before installing libraries
• no I2C backpack on LCD display increased code and wiring complexity