Skip to content

Team · Team Project | Applied Object Detection

Roadside Object Safety Detection

A safety-focused object detection system designed to spot roadside hazards and near-vehicle obstacles that drivers frequently overlook — scoped as assistive perception research, not certified ADAS.

My contribution

I contributed to detection-focused experimentation and framing outputs as awareness support. Scope stayed research/prototype, not certified vehicle software.

Python · Computer vision · Object detection · YOLO · OpenCV · Deep learning · Real-time inference

Problem

Blind spots, roadside encroachments, and sudden peripheral obstacles pose severe collision risks in cluttered driving environments.

Solution & architecture

Built a low-latency visual perception pipeline tailored for forward and peripheral road views, prioritising high-recall detection of vulnerable road users, debris, and roadside hazards.

Perception focus & safety framing

Emphasis on small-object sensitivity, edge-case occlusion handling, and real-time frame throughput — prioritising recall on critical hazard classes while avoiding production ADAS claims.

My contribution

Team contribution to detection work and responsible safety framing for prototype outputs.

Impact

Provides an assistive perception layer aimed at reducing reaction time and preventing near-vehicle collisions in research and prototype contexts.