Computer Vision for Vehicle Queue Counts

A car-wash operator needed to show customers how many vehicles were waiting at each location. The existing approach exposed camera views; the useful customer information was the queue length.

My contribution I helped develop a computer-vision system using a deep-learning model to count vehicles in the queue. We also built tools for adjusting the algorithm to individual facilities and incorporating new examples as conditions changed.

Engineering focus Sites differed in camera type, image quality, viewing angle and weather. The solution therefore needed location-specific configuration alongside a shared recognition system. Queue counts were made available through an API for display on the customer-facing website.

Delivered capability The system was implemented at several locations and supplied vehicle counts for customers to review. Its public output focused on queue information, supporting a more privacy-conscious way to communicate waiting conditions.