How to mine the value in camera data
Personal image video data, this amount of data is also quite large, and everyone is relevant. We went to a good place, good scenery, it doesn’t matter if you look at it, you must let the camera look. There are also data from all walks of life, drone data, industrial data, medical data, sports and entertainment, news and more. These large amounts of data have developed a big step forward in technology, can their value be fully exploited?
When we deal with such data, it will be a problem with visual big data. Its characteristics are obvious. The first is that the amount of data is very large. For example, to process tens of thousands of cameras, the efficiency of calculation is very high. Like city computing, the place with the largest amount of visual data is in the city.
There are some TV stations with 1 million hours of data, which is already a lot. Later, if you have 100,000 cameras in a city, it will take 1 million hours to run for 10 hours. Second, many applications have real-time requirements. For example, adaptive optimization of traffic traffic lights requires real-time analysis and decision making in real time. The third point is that the complexity of the data is very high, the data in all cases, the data of various applications, the cleanliness and quality of the data are very different, the tasks that need to be completed, the intelligence of development They are all different, and the requirements for the universality of the algorithm are very high.