Intelligent Network Surveillance Camera Applications Upgraded to Focus on Each

Compared with front-end intelligence, which is the intelligent application of network cameras, there is another application form in smart applications that is center intelligence, also called back-end intelligence. Unlike the front-end intelligence that uses the front-end camera's own chip to perform calculations, the central intelligence uses the software of the back-end computer to perform the analysis. That is, the camera does not provide any intelligent analysis and operation results, but simply transmits the image information to the management center. All intelligent operations and analysis are performed by the center.

In addition to the central intelligence solution, the management center does not need any analysis of the image because the intelligent analysis and calculation software with perfect performance is configured to form three modules: central intelligent image recognition, client intelligent image recognition, and third-party data analysis. All image data must be transmitted to the management center. Therefore, the center often needs one or more management servers with large capacity and superior performance.

Front-end intelligence and central intelligence each have their own advantages Front-end intelligence and central intelligence cause differences in application performance due to differences in their architectural solutions. First, the front-end intelligence has three advantages in terms of performance. First, the real-time performance is high. The camera immediately analyzes and reports the collected video source. This can effectively improve the system's operating efficiency. Second, the camera first performs the video image. Analysis and identification can only transmit abnormal images to the center for recording and display. These images generally only take up to one-tenth of the total image data, which can greatly reduce the network bandwidth pressure and ensure the efficiency of the entire system. Third, because video compression often loses part of the real information and generates some noise signals. If the system analyzes and recognizes the compressed video, it can easily result in missing reports or false positives. The front-end intelligence is generally in the process. All of these are based on the original video that has been collected and uncompressed for analysis and calculation. This effectively eliminates the above problems. However, the front-end intelligence also faces some technical defects. The most significant is the performance and capacity of the current front-end DSP chip. Compared with the back-end processing capability of large-scale servers and computer software, the front-end intelligence is often used in some intelligent analysis. In high-end industries, such as public security, due to the complex and diverse requirements for intelligent performance, the amount of computation for video analysis is very large. Most DSP chips do not have sufficient memory and processing power to support them. The low processing power will inevitably cause analytical applications. The diversity, as well as the low accuracy of the analysis, led to an increase in false positives. Second, the front-end intelligence is less scalable and flexible, which means that once the system needs to increase or decrease the number of smart cameras or adjust the position, it will be very troublesome to operate.

On the contrary, on the one hand, the central intelligence can achieve very perfect and excellent intelligent analysis performance on the strength of the server's powerful computing processing capabilities, and it is also very convenient for the scheduling and management of front-end resources. On the other hand, as mentioned above, due to its complete reliance on the center, the system must be equipped with a very complete transmission network and server resources in order to deal with the transmission and management of massive image data, which will inevitably bring great cost to the construction. challenge. In addition, in terms of real-time performance and accuracy, the central intelligence model also shows certain disadvantages.

The application of front-end and central intelligence has its own emphasis. First, the actual application of intelligent analysis is still based on high-end industries. Based on the characteristics of front-end intelligence and central intelligence in terms of performance, we believe that the former is more applicable to specific behavior analysis sites. Center intelligence is more suitable for comprehensive behavior analysis and historical data analysis. This has already been demonstrated in the current practical application of traffic monitoring, safe cities, and regulatory industries.

1. Smart applications in the field of traffic jams tend to be front-end prevention The prevention of road traffic violations, traffic accidents, and robbery of motor vehicles is the main purpose of the construction of a traffic bay monitoring system. The core performance requirements mainly include vehicle capture. , license plate recognition, driver recognition, etc., and these properties must rely on intelligent video analysis to achieve, at the same time, this is the relatively high-end requirements of current intelligent analysis applications. From the perspective of intelligent application requirements in this area, this type of demand is more specific and real-time, so it can be completed by the current smart camera.

2. Safe cities and regulatory industries are more suitable for central smart cities and supervisory industries Although there are differences in the manifestations of smart monitoring application requirements, they can be classified as one major category. For example, the safe city first needs to face complex monitoring sites: roads, squares, entertainment venues, etc. Secondly, from the performance requirements, the smart monitoring of Pingan City mainly includes personnel/items detention, abnormal behavior, illegal aggregation, perimeter invasion, etc.; In addition, due to the large number of video resources in Ping'an City, this will inevitably lead to lack of real-time monitoring. Therefore, it is more urgent for smart applications such as quick analysis and retrieval of historical data. From this it can be seen that the intelligent monitoring of Pingan City shows strong complexity, comprehensiveness, and traceability in terms of location and performance requirements, which puts higher requirements on platforms that carry intelligent computing. For current smart camera chip processing capabilities, it is still relatively reluctant. Moreover, the complexity of the monitoring sites also determines that the system must have strong flexibility and scalability in the front-end links. This is also the disadvantage of the front-end intelligence mentioned in the previous article.

The same is true of prisons and detention centers. Although there are no complicated sites in the monitoring system of the supervisory industry, the diversified needs of illegally gathering people, abnormal behaviors, and illegal stay also show a series of complex and comprehensive qualities. Therefore, judging from the current actual cases, the preference for adopting a central smart solution is obviously much higher.

Concluding remarks Of course, both in terms of technical characteristics or market demand characteristics, front-end intelligence and central intelligence are not incompatible with each other, even with each other, complement each other, and a set of video surveillance systems are often not either or both, but must be Front-end and center intelligence are used together. Even in the future, with the continuous improvement of chip technology processing capabilities and the continuous improvement of intelligent algorithms, the market prospects of smart network cameras have been greatly improved. It is believed that the situation where both are coexistent and each has its own emphasis will remain the same.