Co2 Laser Glass Etching,Co2 Laser Glass Cutting,Laser Etched Glass,Laser Crystal Engraving Shandong Meiman Laser Technology Co., Ltd. , https://www.meimanlasers.com
Intelligent monitoring is the combination of intelligent image analysis and video surveillance, so that the monitoring system can detect some events with fixed rules, to a certain extent, reduce the manpower operation. For example, the most common perimeter smart application, a sensitive area is set in the camera screen through the exclusion zone or mix line. Once a moving object enters the area, an alarm will be triggered and the corresponding camera will pop up on the large screen in real time through the platform linkage. Image, alert lights and bells to remind security personnel to handle the incident the first time. Different intelligent applications have basically the same alarm processing after detecting anomalous events. The main difference lies in the image intelligence analysis. Different image intelligence analysis has different requirements, and gradually forms the back-end intelligence and the front-end intelligence. class.
In fact, in the development of video surveillance, there are two aspects of rapid development is recognized by everyone: First, the definition, from the standard era of CIF, D1 into the high-definition era based on 720P, 1080P-based, The camera's image quality has been greatly improved; the second is the system scale, based on mature Ethernet technology, the construction of large-scale digital video surveillance systems has become very simple, with hundreds of front-end system has been everywhere Even large-scale surveillance systems with tens of thousands of front-end points are no longer a fantasy. With the increase in the number of front-end video surveillance systems, it has become impossible to decode all the cameras' images on the wall in the surveillance room. For security personnel, how to discover the events that are happening in the camera images at the first time? It has become a headache. With the image quality getting better and better, can we let the camera understand the events that are happening and raise alarms? This is exactly the reason why the intelligent demand has been very popular in recent years.
Front-end intelligence
With the development of technology, some intelligent analysis from the front end of the back-end server to the front-end camera, through the extra computing resources of the camera, analyze the collected images and extract the key information. The type of pre-configured intelligent analysis is mainly the part of the back-end intelligence that focuses on “live alarmâ€. This part of the analysis requires very few system resources. The front-end camera is sufficient, and the front-end camera can also bring four. Benefits.
1. Save network bandwidth and resolve server bottlenecks. The back-end server must be extra for live analysis
To obtain a real-time video stream, regardless of whether the video stream is directly obtained from the front-end or forwarded from the streaming server, it will cause extra bandwidth consumption in the front-end or back-end, and the server will increase as the number of front-end points to be analyzed increases. The consumption of computing resources will increase and bandwidth consumption will become higher and higher, making it easy for servers to become bottlenecks. After the intelligent pre-configuration, the camera only needs to additionally transmit the analysis result to the background. Compared to the video stream, the bandwidth occupied by the transmitted data is almost negligible.
2. Reduce system construction costs. First, the back-end smart analytics server has added a cost, followed by the back-end
The common intelligent analysis server concurrent processing capacity is about 30 channels D1, 20 channels 720P, and 10 channels 1080P. If you want to increase the concurrent processing capacity, either increase the server configuration or increase the number of servers, but no matter which kind of system will make the system construction cost continue Promote. With intelligent front-end, the cost of a single camera has barely increased, but server-related costs can be greatly reduced.
3, more accurate analysis. The front-end image is transmitted to the back-end server through the network after encoding, and the back-end server solves
Code analysis, encoding, decoding, and possible packet loss in network transmission during this process will affect the accuracy of image analysis. However, intelligent front-end analysis directly in the local, the result will be more accurate.
4, more scalable. In a project, cameras at different points need different intelligent analysis, such as
There is a need for cross-border alarm analysis near the perimeter, face capture at the entrance and exit, license plate recognition on the road, etc. Smart pre-configuration makes it very convenient to configure these intelligent analysis applications by simply replacing the corresponding software version or enabling the corresponding software function. Moreover, special intelligence applications in some special industries can open the interface by professional people for algorithm development. Compared with the back-end intelligent scalability, it is greatly improved.
While seeing the smart benefits of the front-end, we can't ignore the immature nature of the current front-end intelligence, such as low accuracy and weak environmental adaptability. Although in many projects users deploy smart front-end devices, they are too high in practical applications. The false alarm rate and missed report rate caused security personnel to make efforts and finally had to give up.
Such as the most common perimeter prevention application, when an object passes through the lines set in the screen, the camera through image analysis can not accurately determine whether the crossing object is human or animal, which leads to false alarm rate; some cameras reduce by some conditions False alarm rate, such as setting an object larger than 100 pixels to trigger an alarm across the mix line, but because the camera image is near far, the intruder in the distance or the invader who is short and limping will not trigger an alarm. That is, the false positive rate will drop while the false negative rate will increase. The adaptability of the environment, especially the ability to adapt to light changes, has always been a shortcoming of intelligent analysis. Intelligent analysis with high accuracy during the day will be basically unavailable at night.
In fact, these problems are mainly caused by the maturity of intelligent algorithms, and the performance of the main devices such as the camera chip and the Sensor. With the development of technology, problems such as low accuracy and weak environmental adaptability will be solved. Recently, with the launch of the Starlight camera, intelligent analysis under low light has been greatly developed. For example, the HIC5421HI introduced by Yushi Technology is a star-class smart camera that fully considers various lighting mutations, harsh climate and other use environments. Under the low light, it can also provide high-accuracy intelligent applications such as mixing lines, forbidden areas, face capture, and the like, and can also resist the disturbance caused by leaf shaking, shadow changes, and the like.
Industrial Implications of Smart Prepositioning
As more and more smart camera products can be seen, smart pre-positioning is currently a major trend, and it will also have a huge impact on the video surveillance industry. The author believes that the biggest one is openness.
Openness is related to the extensibility mentioned in the previous section. The smart front end provides an open interface, and professional people are responsible for the development of intelligent algorithms to bring about better smart applications. The separation of hardware and algorithms will bring a whole new branch to the video surveillance industry - algorithm providers.
This is similar to the current smart phone industry, different brands of smart phones are equivalent to different manufacturers of smart cameras, all kinds of APP applications are equivalent to different intelligent algorithms, users can choose according to their own needs to install "microblogging" or "Taobao" Even for the same type of intelligent algorithm, the user can also choose among applications such as “WeChatâ€, “Easy Letterâ€, and “QQâ€. From the above, the benefits of separation of hardware and algorithms can also be clearly seen: more user-selectable options, greater flexibility in configuring smart applications, and industry-wide competition to promote smarter maturity.
The difference is incompatible with several smart operating systems in the smartphone industry. The video surveillance industry must have a unified standard. This includes not only the standards between camera hardware and smart applications, but also the structured data standards provided by smart cameras. With intelligent pre-configuration, back-end cloud computing and big data applications can directly use the structured data provided by front-end cameras, which is more efficient. However, if there is no unified standard, various manufacturers set up barriers through private agreements. Greatly hinder the development of intelligence, intelligent front-end is just a piece of paper. Therefore, standardization is another major influence brought by smart pre-positioning.
It is believed that smart pre-location will bring a new look to the traditional security industry, just as smart phones bring great changes to the traditional mobile phone industry. Of course, from the current user feedback on the use of smart front-ends, the current ease of use is not high. It also takes time to gradually improve. We are currently in an era that is equivalent to the Symbian era of smartphones. When will Android and iOS appear? When is the era of truly intelligent pre-production?