From Dashcams to Safety Intelligence: Building a Real-Time Computer Vision System for Unsafe Parking Detection at Fleet Scale

This talk explores how real-time computer vision can be used to improve road safety across large commercial fleets. Using the development of an unsafe parking detection system as a case study, the session covers how dashcam video, vehicle telemetry, and contextual road information were combined to identify high-risk roadside stopping events automatically. The presentation discusses dataset creation, model development, evaluation strategies, edge deployment challenges, and the trade-offs involved in balancing precision and recall in production environments. Attendees will gain practical insights into designing computer vision systems that operate reliably at scale, as well as actionable lessons learned from deploying AI-powered safety solutions across more than 20,000 vehicles.
Format: TECHNICAL TALK, BUSINESS TALK
Location: ROOM B
Speakers: Arzaan Ul Mairaj
Level: Intermediate to Advanced
Audience: Registered participants