Recently, Zhuhai Jinwan has connected the real-time monitoring video resources of water projects such as the coastal river estuary to the AI environmental protection supervision and identification system. In the command hall, surveillance video images of sluice gates and flood discharge channels from Zhuhai Jinwan are being displayed on the large electronic screen in turn. Whenever there are floating objects on the water in the picture, whether it is plastic bottles, foam boxes, or tree stumps, the system will automatically pop up a red box on the target and mark them one by one. This is equivalent to realizing the "automatic watch" of the key nodes of the surface water environment, and installing a "smart eye" to ensure the safety of the water environment. Recently, Zhuhai Jinwan introduced new measures in AI technology to help environmental protection supervision, and connected the real-time monitoring video resources of water projects such as coastal river estuary to the AI identification system. This is also another functional expansion of the system by Zhuhai Jinwan after the launch of the first pollution prevention and control pilot project in Zhuhai using public video resources and AI (Artificial Intelligence) video image recognition technology in June last year, providing scientific assistance for the modernization of the ecological environment governance system and governance capacity. Nowadays, technology enables ecological environmental protection, which is enabling functional departments to continuously establish a more accurate and efficient regulatory system, jointly safeguard ecological security, and constantly polish the gold lettered signboard of Zhuhai's ecological environment. All weather video monitoring and intelligent identification of foreign matters in Jinwan waters. The relevant artificial intelligence monitoring technology service project is the latest achievement of Zhuhai's exploration and development in water ecological security, water environment protection and water environment governance. From manual irregular inspection to electronic continuous monitoring, from human eye judgment to system identification, the use of the new platform has greatly improved the efficiency of water environment supervision. "Through the AI intelligent algorithm of the video system, the surveillance camera has the ability to actively identify environmental pollution problems and hidden dangers such as floating objects on the water surface and water color changes, and automatically generate evidence information such as evidence chains, videos, screenshots, etc." The relevant person in charge of Jinwan branch of Zhuhai ecological environment bureau introduced. In short, this project is to build various types of urban video surveillance for AI supervision, such as coastline gates, key rivers, urban roads and so on, based on the cross departmental sharing of existing video resources in Jinwan District. The system is equipped with automatic identification, automatic alarm, violation review and other functions, combined with manual review of results and further analysis and troubleshooting, to improve the efficiency of environmental supervision by improving the ability to perceive the ecological environment. From the "environmental smart eye cockpit" operation interface displayed on the platform, you can not only clearly see the relevant records of each video recognition water environment event, but also show the overall characteristics of the past period through the analysis of historical records. Thus, relevant supervisors can summarize the rules and make appropriate management service adjustments. "Through this real-time feedback and the analysis of big data behind it, it will help to break the unfavorable situation of lagging and passively dealing with relevant environmental events in the past, and form a good effect of early grasp and preventive management." Introduction of the relevant person in charge of the project technical team. In terms of effect, after the platform was launched on April 29 this year, it automatically identified environmental pollution problems and hidden dangers by accessing 50 channel river sluice monitoring videos and using algorithms. At present, environmental pollution problems and hidden dangers have been detected and identified at 18 monitoring points (50 channel monitoring video stream), and a total of 4049 water pollution risk factors have been identified, most of which are floating debris pollution, in addition to abnormal personnel activities, water quality turbidity and other events. Next, Jinwan branch of Zhuhai Ecological Environment Bureau will continue to promote project optimization and application, and cooperate with technology suppliers to optimize the accuracy of algorithm identification. For example, the platform can learn and upgrade through the accumulated supervision data, strengthen recognition training in the rich data information in actual combat, and improve the application efficiency of system achievements in the continuous "self evolution" of the system, so as to cope with richer intelligent supervision scenarios in the future. Compared with manual inspection, the effectiveness of law enforcement has been greatly improved. The application of AI "environmental protection smart eye" to the prevention and control of surface water pollution is in fact a continuous extension of the working idea of expanding public video resources +ai algorithm. As early as last June, Jinwan branch of Zhuhai Ecological Environment Bureau launched the first pollution prevention and control pilot project in Jinwan District in Zhuhai, which uses public video resources to implant AI (Artificial Intelligence) video image recognition technology. The project has built an environmental supervision platform with the main function of capturing and identifying environmental pollution events such as black smoke vehicles and mud trucks not covered. "Since the project has been running for one year, it can be said that the operation effect is far better than expected." The relevant person in charge of Jinwan branch of Zhuhai Ecological Environment Bureau said that through data statistical analysis, information sharing between departments, and the combination of interview supervision, exposure warning, education and guidance, the platform has solved a number of outstanding problems such as vehicle pollution in the case of "zero fines". The data comparison is the most convincing. In the past year, the pilot has only tested 10 cameras and captured a total of 3297 suspected environmental violations. In the same year, four people and two vehicles dispatched according to the traditional inspection and law enforcement method in the past found 705 problem points. In contrast, the effectiveness of AI "environmental protection smart eye" was as high as 4.6 times. Especially in the past, the law enforcement team composed of multiple departments carried out joint road inspection and law enforcement on black smoke vehicles on the main roads, often with large investment but little effect. Even if individual illegal vehicles are intercepted, the team of their organization will consciously avoid it after knowing it, which is not conducive to the fixation of law enforcement evidence. By means of "public video resources +ai video image analysis", 24 × 7. All weather and all-time supervision. Holidays and nights are no longer "blind spots" that are difficult to supervise. Through the mode innovation of "data sharing" and the technological innovation of "intelligent algorithm", we can realize the substantial "value-added" of low-cost and short-term data. For environmental violations, "we should see clearly and grasp accurately" is the means, and the purpose is to "reduce the number of violations and improve the quality of the environment". On this basis, the data collected by the whole platform are analyzed, and the application of expanding its achievements is also being carried out. For example, according to the recognized license plate number, the ecological environment department can trace the source to the major vehicle users, and conduct interviews, education and rectification related to the environmental protection responsibility. The key link in the management of large car users has been grasped, and the environmental pollution incidents have been significantly improved. From November 2021 to April 2022, there was no major vehicle pollution incident in Jinwan for six consecutive months. In resolving environmental risks, enterprises also have a "Health Code", which is similar to the health code under the current epidemic prevention and control. The "enterprise environmental protection and health code" of Zhuhai Jinwan was launched in May this year. Enterprises can see at a glance whether the online monitoring data is normal, and pay attention to and rectify to eliminate subtle risks with timely prompts. At the same time, the system can also record the environmental protection behavior and pollution status of enterprises in the whole process, and the real-time "chain lock" cannot be tampered with, so as to promote the deep integration of pollution data with the behavior of enterprises and regulators. It is reported that the "environmental protection and health code" is to upgrade the original volatile organic compounds and odor gas precise control system, collect high-frequency and dynamic updated enterprise exhaust emissions and production status data in real time, and use intelligent data analysis to generate color two-dimensional code form. The "enterprise environmental protection and health code" makes complex environmental monitoring data intuitive and real-time, and the ecological environment department can accurately give enterprises the necessary reminders and help guidance. The system intelligently analyzes and discolors nearly hundreds of thousands of data generated by the front-end monitoring equipment every day, feeds back to the enterprise at the first time, points out the pollution risk, tracks the hidden danger and implements the rectification. This measure also reflects the whole process of "pollution production, pollution control and pollution discharge". The health code is not only a simple judgment of conventional data such as emission concentration, but also integrates the current, temperature, flow rate and other information of the pollution production link (production condition) and pollution control link (pollution control facility condition), which can realize the judgment management of the whole process. It can not only reflect the problems more comprehensively, but also give a more accurate prompt of possible problems, and reduce the pressure of enterprise troubleshooting and rectification by improving "pertinence". The application of health code effectively optimizes the early warning mode and reduces the management pressure. In the past, it was necessary to arrange professional environmental protection technicians to spend more than 5 hours a day reviewing nearly hundreds of thousands of data. In the case of no manual omission of review and judgment, the identification and supervision reminder of the next step can be carried out only after all data are reviewed. Now we use "computing power" instead of "human power" to realize real-time automatic code assignment and color change. Compared with the original daily manual review, feedback and reminder, we can achieve "real-time" early warning, and greatly improve the sensitivity of enterprises to abnormal data. At the same time, the health code can understand the conclusion without professional knowledge, making the whole process more convenient and fast. Through the "Trinity" monitoring system data platform, the regulatory authorities can view the probability of different health codes of each enterprise, making it more intuitive to view the ranking of each enterprise, which is conducive to the regulatory authorities' hierarchical control and precise supervision of enterprises. High "green code rate" enterprises holding "environmental protection and health code" can be included in the "positive list" management, and the regulatory authorities will "do nothing without interference". (Xinhua News Agency)