Mexico. In 2023, more than 365 thousand traffic accidents were recorded in urban and suburban areas of Mexico, according to INEGI figures. Against this backdrop, Hikvision presented a new generation of cameras and servers with large-scale artificial intelligence, aimed at vehicle control and incident detection.
AI models trained on real-world situations
Hikvision's large-scale traffic models are based on the Guanlan platform (Hik AI). They were trained with millions of situations such as wet roads, construction zones, and extreme weather conditions. They use the Transformer architecture to simultaneously analyze several parts of the data and apply self-attention mechanisms that allow key information to be extracted considering the overall context of the image, reducing false detections.
Vehicle control cameras
The iDS-TCV9(C)00-HI and iDS-TCV9(C)07-HIR cameras incorporate these AI models for vehicle compliance monitoring, such as detecting driver seat belt and cell phone use.
In the case of the seat belt, the system evaluates the driver's posture and the visible parts of the belt, even if it is partially covered, and reduces false alarms by 75%. To detect phone use, the cameras analyze factors such as hand position, near-ear area, and line of sight, which also reduces errors in identifying this violation by 75%.
Cameras and servers for incident detection
Large-scale AI models are also present in the iDS-TCS802-HS, iDS-TCS907-HIR, iDS-TCS917-HI and iDS-TCS900-HI cameras, as well as the iDS-TSS500-H/16 and iDS-TSS300-H/04 servers. This equipment reduces false alarms by 60% due to incidents such as fallen objects, the presence of pedestrians or stopped vehicles.
The cameras better interpret the context of the image and recognize specific signals such as reflective vests or hazard lights, facilitating more accurate classification of events. Servers allow you to take advantage of these features without the need to renew your existing infrastructure.
"These new cameras represent a significant advancement in traffic management efficiency," said Hollis Li, product manager for smart traffic at Hikvision. "By integrating AI models on a large scale, we offer more accurate solutions adapted to real contexts, which directly contributes to safer, smoother and more sustainable urban mobility."
