The invention discloses a railway crossing
pedestrian and traffic safety detection
system and equipment based on
machine vision, relates to the technical field of
machine vision, and aims to solve the problems of many potential safety hazards, high cost and low efficiency of traditional manual guarding. The
system comprises an
image acquisition module (a crossing south-north high-definition camera, the resolution is greater than or equal to 1920 * 1080), an intelligent analysis module (an AI
edge server, a deployment YOLOv11 model, and the mAP is greater than or equal to 90%), a
signal detection module (a
pressure sensor 1000 meters away from the crossing), an alarm control module (a PLC, an audible and visual alarm and the like) and a centralized control center module, and a monitoring area is divided into a forbidden area, a dangerous area and a waiting area. When a
train approaches, multi-level early warning is triggered, and when abnormity (for example, a forbidden area target stays for more than 3 seconds), second-level alarm is triggered and transmitted to a centralized control center, and a fault-tolerant mechanism is further arranged. Equipment comprises acquisition, analysis, communication and storage equipment, unattended operation of the crossing is realized, the accident rate and cost are reduced, and the
traffic efficiency is improved.