Embodiments of this software relate to the sphere of computer visual technologies, and specifically, to a target tracking method and apparatus, a computer storage medium, and a machine. Target monitoring is among the hotspots in the field of laptop imaginative and prescient research. Target monitoring is broadly utilized in a plurality of fields such as video surveillance, navigation, army, human-pc interplay, virtual reality, and autonomous driving. Simply put, goal monitoring is to investigate and monitor a given target in a video to determine a precise location of the target in the video. Embodiments of this utility provide a target monitoring technique and ItagPro apparatus, a medium, ItagPro and a gadget, to successfully forestall occurrence of cases similar to shedding a monitoring goal and a tracking drift, to ensure the accuracy of goal monitoring. FIG. 1 is a schematic diagram of an software situation of a goal tracking method based on an embodiment of this software. FIG. 2 is a schematic flowchart of a goal tracking method in accordance with an embodiment of this application.

FIG. 11 is a schematic structural diagram of one other goal tracking apparatus according to an embodiment of the present application. FIG. 12 is a schematic structural diagram of a target tracking device in line with an embodiment of this utility. FIG. 13 is a schematic structural diagram of another target tracking device based on an embodiment of this application. Features are often numeric, however structural options resembling strings and graphs are used in syntactic sample recognition. Web server. During actual software deployment, the server could also be an impartial server, or a cluster server. The server may concurrently provide target monitoring providers for a plurality of terminal units. FIG. 1 is a schematic diagram of an software scenario of a target monitoring methodology according to an embodiment of this utility. One hundred and itagpro device one and ItagPro a server 102 . 101 is configured to send a video stream recorded by the surveillance digicam a hundred and one to the server 102 .

102 is configured to carry out the target tracking methodology supplied in this embodiment of this application, to perform target tracking in video frames included in the video stream despatched by the surveillance digicam one hundred and one . 102 retrieves the video stream shot by the surveillance camera a hundred and one , and performs the next information processing for each video frame in the video stream: the server 102 first performs detection in an overall range of a current video body through the use of a goal detection model, to obtain all candidate regions current in the video frame; the server 102 then extracts deep options respectively corresponding to all of the candidate areas in the current video frame through the use of a function extraction mannequin, and calculates a characteristic similarity corresponding to every candidate area according to the deep feature corresponding to the each candidate area and a deep characteristic of the target detected in a earlier video frame; and the server 102 further determines, in line with the feature similarity corresponding to the each candidate region, the goal detected within the previous video body.

102 first performs target detection in the overall range of the current video body by utilizing the target detection mannequin, iTagPro smart device to find out all of the candidate regions present in the present video frame, after which performs target monitoring primarily based on all the determined candidate areas, thereby enlarging a target monitoring range in each video body, in order that occurrence of a case of dropping a monitoring goal due to excessively quick movement of the monitoring goal can be effectively prevented. 102 additionally extracts the deep options of the candidate regions by utilizing the function extraction mannequin, and determines the monitoring target in the present video body based mostly on the deep options of the candidate areas and the deep function of the goal detected in the earlier video body. Therefore, performing target tracking based on the deep function can ensure that the determined tracking target is extra correct, and ItagPro effectively stop a case of a monitoring drift.

FIG. 1 is just an example. FIG. 2 is a schematic flowchart of a goal tracking methodology based on an embodiment of this application. It is to be understood that the execution body of the target monitoring method will not be limited only to a server, but in addition could also be applied to a machine having a picture processing operate corresponding to a terminal device. When the server must perform target tracking for a first video stream, the server obtains the primary video stream, and performs an information processing process shown in FIG. 2 for a video body in the first video stream, to trace a target in the primary video stream. Further, the data processing process shown in FIG. 2 is carried out for a video body in the obtained first video stream, to implement goal monitoring in the first video stream. FIG. 2 for a video body in the primary video stream, to implement goal tracking in the primary video stream.