Answers are grounded in this paper and cite their sources.
Real-time identification of construction workers and compliance verification under dynamic drone surveillance face severe bottlenecks, primarily due to the vulnerability of standard QR codes to motion blur, acute viewing angles, and low resolution under full-coverage personal protective equipment (PPE). To address this gap, this study proposes an optimized, detector-free video processing pipeline featuring independent non-blocking stages, RTSP/WebRTC streaming, and Redis-backed temporal deduplication to evaluate fiducial AprilTags versus QR codes across varying badge heights and tracking conditions. Empirical findings demonstrate that AprilTag achieves a superior decode yield of 0.623 in detector-free configurations. It outperforms QR codes (0.161) by 3.9× and maintains operational viability at minimum badge heights of 24–32 pixels, whereas QR codes fail below 48 pixels. Furthermore, track-level temporal accumulation reveals that AprilTag requires only 4 sightings to attain 95% worker identification confidence, compared to 18 sightings for QR codes, while ensuring low-latency streaming and efficient storage control. These results establish robust camera placement specifications and track-level accumulation as core metrics for real-industrial video surveillance. Future research will investigate multi-drone orchestration and adaptive image enhancement techniques.
No takes yet. Share an insight, caveat, or question.
Омаров et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: