Answers are grounded in this paper and cite their sources.
Street lighting accounts for a significant share of municipal electricity consumption, and most installations still operate at full brightness from dusk to dawn regardless of road activity. This paper presents a smart street light system that adjusts lamp intensity according to real-time road activity. Video frames from a roadside camera are processed by a lightweight deep-learning object detector to recognize pedestrians and vehicles; a tracking module assigns persistent identities and estimates movement direction; and a rule-based controller converts the detected activity into a pulse-widthmodulation (PWM) dimming command for an LED driver. When the road is empty, the lamp holds a low baseline level that satisfies minimum safety requirements, and it ramps up when a pedestrian or vehicle is detected in, or approaching, its zone. Two candidate designs were compared, and a detector-plus-tracker pipeline was selected for its balance of accuracy and real-time performance on edge hardware. The selected design reaches 92.6 percent precision and 89.4 percent recall with 34.2 ms inference latency, and the estimated energy saving over constant-output lighting ranges from 34.5 to 65.6 percent depending on traffic density. The work shows that vision-driven adaptive lighting can reduce energy use while preserving road safety and provides a scalable basis for smart-city lighting.
No takes yet. Share an insight, caveat, or question.
Kumar et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: