TS processing method friendly to artificial intelligence algorithm and application thereof

Abstract

The invention relates to processing of video stream, and especially relates to a TS (Transport Stream) processing method friendly to an artificial intelligence algorithm and an application thereof. The TS processing method comprises the steps of: (1) receiving a TS; (2) judging whether there is a need to add a data packet of structural data and / or business rules; if it is judged that there is no need, continuing receiving the TS; and if it is judged that there is a need to add the data packet of the structural data and / or business rules, performing operation of a step (3); and (3) segmenting the data packet of structural data and / or business rules into at least one data packet in the form of Null Packets, and continuously inserting the data packet to a position prior to the current TS. TheTS processing method of the invention has the beneficial effects that: the TS with packaged data completely meets a standard, a third-party player or browser has to discard the data packet in the form of Null Packets according to requirements of a TS standard, therefore, a playing effect of the player can not be influenced by adding of the data packet in the form of Null Packets, at the same time, in the TS standard, through adoption of the Null Packets, data of random lengths can be packaged, therefore, application of the TS processing method can not be limited by specific project requirements, and the data of random lengths can be packaged in the data packet in the form of Null Packets, so as to be provided for use of the artificial intelligence algorithm.

Description

technical field [0001] The invention relates to the processing of video streams, in particular to a TS stream processing method and application friendly to artificial intelligence algorithms. Background technique [0002] At present, artificial intelligence vision algorithms are increasingly being used in large-scale projects such as "smart cities" and "urban brains". In these large-scale projects, there are often multiple algorithm companies involved in visual algorithm processing at different levels or businesses. The video stream from the front-end camera is pushed to the algorithm server of Company A, which analyzes various structured data from the video, such as the location of cars, pedestrians, models, and license plates, and then encapsulates these structured data into the video stream. Continue to push to the algorithm server of company B, and so on. [0003] At the same time, business rules will also be encapsulated in the video stream for use by various algorith...

AI Technical Summary

Problems solved by technology

[0021] It is difficult to accurately synchronize the time of video and data, which will lead to two problems: First, different algorithm servers will adopt different business rules for the same video stream, such as administrators modifying business rules, starting tide lane judgment, algorithm server A After receiving the new rule, the algorithm server B has not received the new rule, which leads to a deviation in judgment; second, because of the delay, when the algorithm server receives the new rule, the video stream is still the scene a few seconds ago. Therefore, after the rule is changed , there will be a few seconds of misjudgment

Images