CCTV Finance's special program "Super Production Line" has just featured Changzhou Haitu Technology, showcasing in a live broadcast how the company empowers urban governance with AI, with simultaneous promotion across CCTV Finance's Weibo, official account, video channel and other new media platforms, drawing over one million views across the internet.
"Hello everyone, I am Zhao Xinyu, product manager from Changzhou Haitu Technology." Standing at the entrance of a workshop corridor, Zhao Xinyu appeared on camera with the "Personnel Overlimit Recognition System" developed by Haitu Technology for dust-explosion-risk workshops behind him.
"Dust-explosion-risk workshops have strict personnel limits. Traditional methods rely on manual inspections and door registration, which are not only inefficient but also prone to missed or erroneous records. Haitu's solution uses AI visual detection to count people entering and exiting through multiple passages in real time, and the system automatically calculates the total number of people in the area. Once the preset limit is exceeded, an on-site alarm is triggered, and safety officers also receive warning messages on their mobile devices," Zhao Xinyu said.
This means safety officers no longer need to repeatedly visit the site to count people, as the system can determine in real time whether an overlimit situation has occurred. CCTV's cameras then switched to Haitu Technology's 7×24-hour safety operations center. The surveillance footage of that dust-explosion-risk workshop was displayed in real time on the large screen of the AI intelligent safety supervision service platform developed by the company.
Xiao Tao, chairman of Haitu Technology, introduced that the platform can be applied to scenarios such as factory-within-factory parks, resettlement housing communities, small public venues, and fireworks and firecracker retail outlets, and already supports on-demand combination of 15 major categories of core functions. Its core technological capability lies in the two-way integration of "AI video" and "IoT sensing." That is, through smart cameras equipped with AI algorithms, the platform can conduct millisecond-level real-time identification and dual warnings for scenarios such as occupied fire lanes, illegal high-altitude work, and e-bikes being brought upstairs in violation of rules. At the same time, through smoke, temperature, and combustible gas sensors, it enables all-weather monitoring of the physical environment. Whether for deep supervision of existing scenarios or the integration of new business formats such as energy storage and warehousing, the system can be rapidly expanded. More critically, the platform adopts a "reuse plus new installation" strategy—making full use of existing monitoring equipment and combining it with newly deployed computing power and algorithms, greatly reducing the cost of digital upgrading.
Because of the massive volume of data, Haitu also developed a "Safety Situation Analysis Intelligent Agent": it aggregates video stream data, warning data, and handling data to conduct comprehensive online analysis and assessment in real time. During the on-site demonstration, someone asked, "How has safety been this month?" The intelligent agent replied that the system monitored a total of 51 alarms in September, up 64.52% year on year, with a significant increase in alarm frequency; however, through AI identifying and notifying on-site handling at the first opportunity, all alarms have now been cleared, and no substantive safety accidents occurred.
Unlike traditional systems that emphasize construction over operations, Haitu places special emphasis on a "7×24-hour operations system." Addressing the previous pain point of alarms going unhandled because duty personnel left their posts, the platform has built a professional online duty team. Once an alarm is triggered, it forms a closed loop of "detection—confirmation—dispatch—handling." In the demonstration, a personnel overlimit warning in a dust-explosion-risk space was reported to the command center immediately, with on-site images and video clips presented simultaneously, and the system tracked when safety management personnel completed the handling.
At present, Haitu Technology's system has been connected to more than 100 factory-within-factory sites, more than 800 small public venues, more than 80 fireworks and firecracker stores, and 10 residential communities, effectively handling more than 1,000 warning incidents. Based on artificial intelligence, Haitu Technology mainly focuses on industrial scenarios, urban renewal scenarios, and the energy sector, aiming to use AI technology to support enterprise safety production and provide solid technological support for safeguarding urban safety and protecting enterprise development.